Choice experiment survey to estimate the economic value of visibility improvement for Canadians
Final Report
Prepared
for Environment and Climate Change Canada
Supplier name: Kantar
Contract number: K1A12-191149/001/CY
Contract value: $122,887.15
Award date: 2019-05-10
Delivery date: 2020-11-30
Registration number: POR # 007-19
Ce rapport est aussi disponible en français
For more information
on this report, please contact ECCC at: ec.enviroinfo.ec@canada.ca
Choice experiment
survey to estimate the economic value of visibility improvement for Canadians Final
Report
Prepared for Environment
and Climate Change Canada
Supplier name: Kantar
November 2020
Environment
and Climate Change Canada (ECCC) commissioned Kantar to design and conduct a
choice experiment survey in order to assess the economic value that Canadians’
associate with a noticeable visibility improvement, expressed in monetary
willingness-to-pay per household for a 1-unit deciview (DV) change. The
findings of this study are meant to refine the accuracy and representativeness
of the economic values associated with visibility in the Air Quality Valuation
Model (AQVM2), whose estimates are used in cost-benefit analyses of air
pollution regulations.
Cette publication est aussi disponible en français sous le titre: Enquête
par la méthode de choix multi-attributs pour estimer la valeur économique d’une
amélioration de la visibilité auprès des Canadiens
Permission to
Reproduce
This publication may be
reproduced for non-commercial purposes only. Prior written permission must be
obtained from ECCC. For more information on this report, please contact ECCC at
ec.enviroinfo.ec@canada.ca or at:
DEPARTMENT
OF THE ENVIRONMENT
200
Sacre-Coeur Blvd.
Gatineau,
Quebec
K1A0H3
Catalogue Number: En4-424/1-2021E-PDF
International
Standard Book Number (ISBN): 978-0-660-37650-9
Related publications
(registration number: POR #007-19):
Catalogue Number: En4-424/1-2021F-PDF (Final Report, French)
ISBN:
978-0-660-37652-3
© Her Majesty the Queen in Right of Canada, as represented by Environment
and Climate Change Canada, 2021
1.1 Research
Purpose and Objectives
2. Economic Value of Visibility Improvement for Canadians
Digital Alteration of the Images
Appendix B: Air Quality Images
Acronym |
Definition |
AAPHI |
Addressing Air Pollution Horizontal Initiative |
AHC |
Annual Household Cost |
AQHI |
The Air Quality Health Index |
AQVM2 |
Air Quality Valuation Model |
CAWI |
Computer Assisted Web Interviewing; the technology used for data
collection |
DV |
Deciview; a measure of visibility
that corresponds to incremental but perceptible changes in visual perception |
ECCC |
Environment and Climate Change Canada |
PNG |
Portable Network Graphic;
the type of photograph output used in the survey |
WTP |
Willingness to Pay |
VAQR |
Visual Air Quality Rating |
Air pollution can lead to haze that can reduce
or obscure visibility and economic literature has established that reduced
visibility can be associated with reduced citizen well-being along with lost
revenues in the areas of outdoor recreation and/or tourism. To improve visibility, it is necessary to
reduce pollution levels, which can come at a cost to Canadian consumers.
Generally, these costs are indirect and come in the form of additional expenses
incurred by businesses for installing pollution control devices on vehicles and
manufacturing equipment. The additional costs to businesses are eventually
passed on to Canadians through higher prices on everyday items such as food,
electricity, and transportation. Reducing pollution and thus improving
visibility means that Canadians will experience unavoidable increases to
general cost of living.
To estimate the value of changes in pollution
levels, Environment and Climate Change Canada (ECCC) currently uses the Air
Quality Valuation Model (AQVM2). This
model measures the impacts of pollution on visibility, crop productivity, and
cleaning costs for households.
The current inputs into the visibility module
within AQVM2 use data that was last collected in 2002 in the lower mainland of
British Columbia only and were applied throughout Canada. Furthermore, the existing empirical
literature pertaining to the valuation of visibility improvement is very
limited, especially in Canada. The
collection of current and more methodologically robust data will allow ECCC to
provide more accurate information to decision-makers, which is consistent with
ECCC’s responsibilities, Treasury Board Secretariat’s guidelines on
cost-benefit analysis under the Cabinet Directive on Regulation and the
Government of Canada’s commitment to evidence-based decision-making.
The overall objective of this research was to
obtain current and robust data regarding Canadians’ willingness to pay (WTP)
for improved visibility that can better characterize the differences that may
exist across the Canadian population.
The findings of this study will be used to refine the accuracy and
representativeness of the economic values associated with visibility in AQVM2,
whose estimates are used in cost-benefit analyses of air pollution regulations.
A discrete choice experiment was undertaken
with the goal of understanding how attributes of visibility, health risk and
annual cost to household affect WTP per household for a 1-unit DV change. The
levels chosen for investigation for this study are outlined in Table 1.2.a.
below.
Visibility (Deciview/Visual Range) |
Health Risk |
Annual Household Cost |
9 DV (155-160 km) 13 DV (105-110 km) 17 DV (70-75 km) 21 DV (45-50 km) 25 DV (30-35 km) 29 DV (20-25 km) 33 DV (10-15 km) |
Low Moderate |
$30 ($2.50 per month) $60 ($5.00 per month) $90 ($7.50 per month) $180 ($15.00 per month) $360 ($30.00 per month) None |
Visual stimuli (pictures) were used to depict
various levels of visibility to the respondents. As there is no “typical” visibility for
Canada, a wide visual range was chosen for testing (5-35 DV) to allow for
evaluation of the most likely air quality scenarios in Canada.
The Air Quality Health Index (AQHI) was used to represent
health risk to respondents and two levels of health risk were included in the
final design: low and moderate. No
constraints were imposed on which health risk levels could be combined with
which visibility levels.
A complete enumeration approach was used while
designing the choice sets. The complete enumeration approach was chosen as it
better addresses the objective of the research: to estimate a robust
nationally-averaged WTP value (annual $ per Canadian household) for a 1-unit
deciview (DV) change and to identify statistically significant variables in
explaining the willingness to pay.
A design with balanced alternative effects
(complete enumeration) does a better job of estimating the specific visibility
levels in the context of the price, whereas a design with imbalanced
alternative effects (full factorial) would be better for estimating the
gaps.
In general, the goal of the experimental design is
two-fold:
1.
Level balance – each level to appear the same number
of times as each other level within an attribute.
2.
Orthogonality – levels across attributes to be
independent of each other in how they appear across choices.
In this study’s design, the following constraints were
implemented with the goal of a more realistic comparison for respondents:
·
For each task, the baseline scenario was on the left
with the test scenario on the right
·
The test scenario always had better visibility than
the baseline scenario.
·
The baseline scenario always had $0 cost
·
The test scenario always had cost of at least $30 per
year.
·
The baseline scenario always had visibility no better
than 17 DV.
·
The test scenario always had visibility no worse than
25 dv.
The discrete choice exercise was estimated
using a Hierarchical Bayes Multinomial Logit model and was estimated using Sawtooth
Software’s CBC Hierarchical Bayes Module v5.5.6. The model used an iterative Monte Carlo
Markov Chain approach to estimate the model for 200,000 iterations with the
first 100,000 iterations used as a burn-in to calibrate the process and the
last 100,000 iterations used to provide a robust estimate of the model. The final model estimated linear effects for visibility
and annual household cost and categorical effects for the two levels of health
risk. This model generated a robust estimate of the WTP per household for each
one-unit decrease in the DV scale for the entire sample and for various
subgroups of interest.
Two WTP values were calculated per
respondent. The first WTP was when the heath
risks are both moderate since we assume the baseline state has a moderate
health risk. In this calculation, the overall utility of the health risk was
zero since both the baseline and improved health risk level were the same. The second WTP value was the WTP for a one
unit decrease in DV that results in a low health risk. This calculation included the change in
utility in moving from a moderate risk to a low risk.
When health risk is zero, on average, Canadians
are willing to pay $107.04 annually or $8.92 per month for an improvement of
one DV to visibility. The median is
$1.10 per month and the standard deviation is $21.27 per month indicating a
wide variability in the amount that Canadians are willing to pay for 1 DV
improvement in visibility.
There are noticeable differences among
different demographic groups, more specifically, younger Canadians (18-34),
households with children or with individuals with health conditions impacted by
air quality and/or those who currently live in areas with high visibility are
all willing to pay more than their respective counterparts.
Not unexpectedly, Canadians are willing to pay
more when there is an associated improvement to health. On
average, Canadians are willing to pay $581.76 annually or $48.48 per month for an improvement of one DV to visibility that includes a
perceived associated decrease in health risk from moderate to low. There are noticeable
differences among different demographic groups when it comes to WTP with an
associated improvement to health. Specifically,
younger Canadians (18-34), women, households with children or with individuals
with health conditions impacted by air quality and/or those living outside of
Atlantic Canada are all willing to pay than their respective counterparts.
In order to provide more clarity around the WTP
differences with improved health risks, the analysis reviewed the ratio of WTP
on its own, compared to WTP with an associated improvement to health risk. Absolute WTP values identify how much
Canadians care about visibility and how much they care about health. The ratio analysis allows one to understand
how much Canadians care about visibility compared with health. Not unexpectedly, virtually all Canadians
care more about health than visibility however, the ratio analysis helps to
identify Canadians that “care” more about visibility and these include middle
aged Canadians (35-54), Atlantic Canadians, rural Canadians, and Canadians with
children in the home and/or living in areas with good visibility (9 DV or less).
The findings of this study are based on online
surveys conducted from September 8 to
29th, 2020. The survey was conducted among Canadians aged
18 years and older. Respondents were randomly selected from an online panel and
invited via email and/or personal online panelist dashboard to participate in
the survey. The results of panel
surveys are considered a non-random sample, meaning they are not a random
selection from the general population of Canada, rather they are a subset of
people who are, in this case, people who have signed up to participate in
online surveys. As such, margin of error does not apply.
The data have been weighted to reflect the
demographic composition of the Canadian population for age, gender, region, education,
and population of residence. Surveying was conducted in the respondent’s
official language of choice and took an average of 15 minutes to complete.
The total contract value for the project was $122,887.15
including applicable taxes.
I hereby certify as a representative of Kantar
that the deliverables fully comply with the Government of Canada political
neutrality requirements outlined in the Communications Policy of the Government
of Canada and Procedures for Planning and Contracting Public Opinion Research.
Specifically, the deliverables do not include information on electoral voting
intentions, political party preferences, standings with the electorate or
ratings of the performance of a political party or its leaders.
Tanya Whitehead
Kantar
Senior Director, Public Practice Leader
To
estimate the value of changes in pollution levels, Environment and Climate
Change Canada (ECCC) currently uses the Air Quality Valuation Model
(AQVM2). This model measures the impacts
of pollution on visibility, crop productivity, and cleaning costs for
households. AQVM2 is also used to generate values that are used for input into
cost-benefit analyses within Regulatory Impact Analysis Statements for
regulations under Addressing Air Pollution Horizontal Initiative (AAPHI), such as the oil and gas sector regulations[1] and the multi-sector air pollutants
regulations[2]. The
main objective of such cost-benefit analyses is to demonstrate the benefits of
air pollution regulations versus their costs.
The
current inputs into the visibility module within AQVM2 use data that was last
collected in 2002 in the lower mainland of British Columbia only and were
applied throughout Canada. Furthermore,
the existing empirical literature pertaining to the valuation of visibility
improvement is very limited, especially in Canada. In fact, at the time of design, the most
recent primary studies in Canada, were published before 2012 or address
historical valuation metrics, and as such it is not possible for ECCC to draw
from a recent study to update the AQVM2 economic values. The collection of
current and more methodologically robust data will allow ECCC to provide more
accurate information to decision-makers, which is consistent with ECCC’s
responsibilities, Treasury Board Secretariat’s guidelines on cost-benefit
analysis under the Cabinet Directive on Regulation and the Government of
Canada’s commitment to evidence-based decision-making.
The overall objective of this research was to
obtain current and robust data regarding Canadians’ willingness to pay (WTP)
for improved visibility that can better characterize the differences that may
exist across the Canadian population. More specifically, this study was designed to
collect the necessary data that will allow ECCC to assess the economic value
that Canadians associate with a noticeable visibility improvement, expressed in
monetary willingness to pay per household for a 1-unit deciview (DV) change.
The
findings of this study will be used to refine the accuracy and
representativeness of the economic values associated with visibility in AQVM2,
whose estimates are used in cost-benefit analyses of air pollution regulations.
A discrete choice
experiment approach was undertaken for this research with the goal of
understanding how attributes of visibility, health risk and annual cost to
household affect willingness to pay per household for a 1-unit DV change. The DV scale is a
visual index designed to be linear with respect to perceived visual changes
over its entire range.[3] The DV scale is
zero for pristine conditions and increases as visibility degrades. A 10%
improvement in visual range (in km) roughly corresponds to a decrease of 1 DV,
regardless of the initial visual range. The
levels chosen for investigation for this study are outlined in Table 2.1.a.
below. More detail on each is provided in their respective sections below.
Visibility (Deciview/Visual Range) |
Health Risk |
Annual Household Cost |
9 DV (155-160 km) 13 DV (105-110 km) 17 DV (70-75 km) 21 DV (45-50 km) 25 DV (30-35 km) 29 DV (20-25 km) 33 DV (10-15 km) |
Low Moderate |
$30 ($2.50 per
month) $60 ($5.00 per
month) $90 ($7.50 per
month) $180 ($15.00 per
month) $360 ($30.00 per
month) None |
This study was designed to use an online panel
sample. While we recognize that the use
of an online panel sample results in a non-random sample, and that mail and
telephone options may address the non-random limitation, we nevertheless found
the online with panel option to be the best design for the objectives of the
study. The design of the study in presenting visual images is not well suited
for a paper-based (mail) or telephone approach and would limit the range of
experimental conditions. As such, the
resulting methodology would require a mail/telephone recruit to online.
Response rates for mail and telephone surveys are very low in Canada[4] and
would result in significant response bias and require large weights to adjust
for response bias. Further, given the WTP
methodology would require a mail/telephone recruit to online approach, one
could argue that the sample was not truly probabilistic as the starting sample
for the survey would be “Canadians who agreed to complete an online
survey”.
The concern generally with non-random sample is
that it may not represent the population. To address the representation concern,
the study included quotas on completions to attempt to ensure our final sample was
representative of the population based on Statistics Canada Census
data. Further, we implemented random selection when matching the
outgoing sample to the population. The online quota sampling approach, while
non-probabilistic in nature, provided the best mode for a willingness to pay
study while continuing to have a sample that is representative of the Canadian
population based on age within gender within region. Details on final
sample composition and weighting adjustments are outlined later in the report.
The sample for this survey was sourced
from Kantar’s Profiles panel. Respondents were invited to participate in the
survey by email and/or personal online dashboard via Profile’s website. Profile’s panel has nearly 100,000 panellists
located across Canada representing every region. The panel includes Canadians who have
opted-in to participate in online surveys, and as per standard requirements
this research excluded panellists who have participated in a Government of
Canada survey or other similar surveys within the past 30 days. Recruitment for the
panel primarily occurs through our longstanding partnerships with large
Canadian brands like Hudson’s Bay Rewards, Aeroplan, Wal-Mart and PETRO-POINTS.
The
original or “base” images were sourced from the Air Quality Science Unit,
Prediction and Services West Division of Meteorological Service of Canada /
Environment and Climate Change Canada.
The original images were used in a public evaluation of the local Visual
Air Quality Rating (VAQR) in 2013. The
images used in this study were chosen for convenience. More specifically, they were images that were
available to the project at no cost, that had a measured DV associated with
them and also included sufficient detail that would make it possible to
perceive changes in visual range in both an urban and rural setting (i.e.,
mountains in the background). Kantar
worked with a variety of scientific experts to ensure the
chosen imagery accurately represented a baseline level of visibility for a
rural and urban location and then engaged a visual design expert to digitally alter
images to depict different levels of visibility for each selected location.
The photo manipulation involved a four-step
process that is described below.
Stage 1 - Base Photo Preparation
Photographs from Burnaby and Chilliwack were
used as base representations of urban and rural environments respectively for
the study. These locations were chosen as these were the locations in which the
Air Quality Science Unit, Prediction and Services West Division of
Meteorological Service of Canada / Environment and Climate Change Canada had a
range of images with confirmed DV measures. A DV 9 photo for the urban and a DV 10 photo
for the rural representation were retouched in Photoshop to remove clouds and
improve contrast. Haze levels were cleared so that each base photo most closely
represents a DV 9 reference photo. Detailed photo manipulation included the
following:
1.
Split
photograph into 5 distinctive layers in Photoshop.
2.
Split each
photo into layered sections to be separately adjustable.
3.
Layers for
rural environment: sky, mountains, nearby hills, farmland, and immediate
foreground.
4.
Layers for
urban environment: sky, mountains, city skyline, neighbourhood, and immediate
foreground.
5.
Remove or
add haze to different layers by adding or lessening contrast to mid-tones in
Adobe Camera Raw.
6.
Define
haze on different layers by adding or lessening texture and clarity in Adobe
Camera Raw.
7.
Adjust
colour of haze on different layers by adjusting exposure, contrast, highlights,
shadows, and colour temperature according to represented DV photos.
Stage 2 – Preparing photos for urban and rural
DV
Next, base 9 DV photos for the urban and rural
areas were first digitally manipulated in Photoshop to correlate to the
following seven (DV) levels 9, 13, 17, 21, 25, 29, and 33. Photos representing DV
levels 9, 10, 11, 12, 13, 15, 16, 22, 23, 24, 28, and 29 along with model
generated photos for DV levels 9, 13, 17, 21, 25, and 33 were used as reference
levels. The 5 layers of base-level photograph were adjusted to accurately
represent the seven levels of haze (DV).
Detailed photo manipulation involved the following:
1.
Adjusting
haze levels by adding or lessening contrast to mid-tones in Adobe Camera Raw.
2.
Further
defining haze level on different layers by adding or lessening texture and
clarity in Adobe Camera Raw.
3.
Adjust
colour of haze on different layers by adjusting exposure, contrast, highlights,
shadows, and colour temperature in Adobe Camera Raw.
4.
Add artificial
haze as needed where Adobe Camera Raw could not compensate.
Stage 3 – Photo retouching revisions
Photos for the digitally altered DV levels of
9, 13, 17, 21, 25, 29, and 33 were reviewed against reference photographs,
combined with project team and external reviewer feedback, and adjusted. Adjustments included refinements for the
environment related to haze levels and adjustments to different layers’ level
of haze by adding or pulling back the transparency of the haze as well as
removing or adding artificial haze.
Stage 4 – Preparation for web survey
The final images were cropped and prepared for
the web survey with the following specifications:
1.
Crop
photos to 500 x 332 pixels
2.
Convert
each photo to 24-bit depth
3.
Finalize
photograph output to be non-transparent, 72dpi PNG (Portable Network Graphic)
Two levels of health
risk were included in the final design: low and moderate. No constraints were imposed on which health
risk levels could be combined with which visibility levels. The Air
Quality Health Index (AQHI) was used to represent health risk to
respondents. AQHI uses an index by
estimating the daily change in mortality risk for ten cities from 1998-2000 and
plotting it on a 10-point scale[5]. The
higher the number, the greater the risk and the need to take precautions. It is
a personal health protection tool for individual Canadians including those at
higher risk and focuses only on health risk, i.e., it does not attempt to
consider any issues other than the day-to-day health impact of air pollution.
High and very high levels on the AQHI were excluded from the final design
after pre-testing found that respondents with high or very high health risks
options were unable to separate the two constructs of health and visibility and
as such were unwilling to consider any payments for scenarios which included
high or very high levels on the AQHI.
AQHI
was chosen for this research as it was developed through a national process and
was designed to apply across the country and uses a scale that is both
continuous and categorical. More
specifically, AQHI provides a number from 1 to 10+ to indicate the level of
health risk associated with air quality and also provides categories that go
with the numbers of low, moderate, high, and very high. The categories
have standardized definitions which are regularly used in the general public and
are outlined in Table 2.2.a. below.
Risk |
Air Quality Health
Index |
Health Messages |
|
At Risk Population* |
General Population |
||
Low |
1-3 |
Enjoy your usual outdoor activities. |
Ideal air quality for outdoor activities. |
Moderate |
4-6 |
Consider reducing or rescheduling strenuous activities outdoors if you are
experiencing symptoms. |
No need to modify your usual outdoor activities unless you experience symptoms
such as coughing and throat irritation. |
High |
7-10 |
Reduce or reschedule strenuous activities outdoors. Children and the
elderly should also take it easy. |
Consider reducing or rescheduling strenuous activities outdoors if you experience
symptoms such as coughing and throat irritation. |
Very High |
10+ |
Avoid strenuous activities outdoors. Children and the elderly should
also avoid outdoor physical exertion. |
Reduce or reschedule strenuous activities outdoors, especially if you experience
symptoms such as coughing and throat irritation. |
The average annual cost to a household required
to achieve the visibility presented was included at 6 levels:
‒ $30 ($2.50 per month)
‒ $60 ($5.00 per month)
‒ $90 ($7.50 per month)
‒ $180 ($15.00 per month)
‒ $360 ($30.00 per month)
‒ None
The cost was presented in both annual amounts
and monthly amounts to ease comprehension to respondents. It was also presented in a manner that would bring
about a permanent improvement in visibility and that the cost was
unavoidable. Specifically, “this cost is
unavoidable – while you would NOT be charged a specific fee or additional tax,
you would experience the cost through increases in your cost of living”. The
payment vehicle was chosen to be most reflective of how the increased costs
would be applied in practice. In this
case the implementation of policies/regulations would not result in increased
taxes to the consumer; rather it may increase production costs, which are often
passed along to consumers.
A
design with balanced alternative effects (complete enumeration) does a better
job of estimating the specific visibility levels in the context of the price
whereas a design with imbalanced alternative effects (full factorial) would be
better for estimating the gaps.
Each
respondent was randomly assigned to one of the design versions. Once the respondent chose one of the two
options, that design version was removed from the set of available versions
until the respondent had been shown the full list. The order of tasks within each version were
randomly ordered. However, the
attributes (visibility, health risk, setting and annual household cost)
remained consistent throughout the activity.
Respondents were exposed to a combination of
the following attributes and levels, with a total of 1,600 possible
permutations.
In general, the goal of the experimental design
was to have level balance, that is, for each level to appear the same number of
times as each other level within an attribute and to have orthogonality, for
levels across attributes to be independent of each other in how they appear
across choices.
However, in this design we had the following
constraints with the goal of more realistic comparisons for respondents:
The questionnaire
was designed to generate estimates of WTP per household for a 1-unit decrease
in DV (roughly equivalent to a 10% improvement in visual range). The design
included control mechanisms such as scripts to minimize respondent biases,
follow up questions to distinguish between legitimate zero willingness to pay
and protest responses and ways to isolate WTP for health-related improvement
from WTP for visibility-related improvement.
The questionnaire
included background knowledge information to the respondent and general understanding
by respondents was confirmed during pretest.
This included scripts and other background information that were
presented upfront to survey respondents in order to reduce known bias risks
that might occur. For example,
participants were asked to read a script that outlined how the implementation
of pollution reducing measures translated into costs for consumers and information
that explained that visibility is not always a good indicator of health risk;
and then answer true and false questions about what they had just read. If they answered incorrectly, they were
looped back to the script and asked to read again. For full details on the survey instrument, see
Appendix A.
A pretest was undertaken in March 2020 that
included 47 completes, of which at least 10 were in French. Minor wording
changes were made to some of the instructions after the pretest and, as such,
the completions were not included in the final data set. In March 2020, the first wave of COVID 19
pandemic hit Canada and fieldwork for this project was placed on hold.
Data collection
was conducted online from September 8 to 29th,
2020. Given the COVID-19 pandemic
was an ongoing factor during fieldwork, there are a number of considerations
when interpreting results. First,
economic uncertainty, high unemployment rate, and possible budget constraints
may have reduced WTP among some respondents.
Conversely, higher interest in outdoor recreational activities and
environmental conservation may have increased WTP among some respondents.
The 15-minute online survey was conducted using
computer assisted web interviewing (CAWI) technology. CAWI ensures the interview flows as it should
with pre-programmed skip patterns. It
also controls responses to ensure appropriate ranges and data validity. Surveys were conducted in English or French
as chosen by the respondent. All
participants were informed of the general purpose of the research, the supplier
and that all of their responses would be confidential. At the end of the survey, respondents were
informed of the sponsor to avoid inducing bias in responses.
Respondents
were randomly selected from Kantar’s online panel and invited to participate in the survey by email and/or personal online dashboard
via Kantar Profile’s website. Panellists
who participate in surveys are incentivized through a points system that is
redeemable for a variety of gift cards. As such points were provided as remuneration
for participating in the survey.
To allow for
robust sub-analyses, a sample of 2,000 Canadians was assembled, with
interlocking quotas on completions for age within gender within region to
ensure the sample was representative of the general Canadian population aged
18+ with a threshold of +/- 5 percent (Table 2.4.a).
Atlantic |
Quebec |
Ontario |
Prairies |
BC |
Totals |
|
Males 18-34 |
17 |
62 |
114 |
59 |
38 |
289 |
Males 35-54 |
21 |
78 |
128 |
63 |
43 |
334 |
Males 55+ |
27 |
89 |
136 |
57 |
51 |
360 |
Females 18-34 |
16 |
60 |
113 |
56 |
37 |
282 |
Females 35-54 |
22 |
76 |
132 |
61 |
45 |
335 |
Females 55+ |
30 |
98 |
154 |
61 |
55 |
400 |
Totals |
134 |
463 |
777 |
358 |
268 |
2000 |
To aim for the sample to be representative of
the Canadian adult population 18+, the design first implemented controls using
quota sampling. Quota variables included
gender, age, and region. In addition to
the previously mentioned quotas, the final sample was weighted using various
demographic information available from Statistics Canada outlined below. Weighting did not include an
income variable because population level data for income was only available at
the household level and weights were being applied at an individual level.
The representativeness of the
sample was validated on 4 dimensions:
1. Region, gender, and age
2. Region and population of residence
3. Region and education
4. Education, gender, and age
The validation of the education,
gender and age dimensions identified low counts for ‘Some high school or less
education’ and “Apprenticeship or other trades” across all gender and age
combinations. As such, these groups were
merged with ‘High school diploma or equivalent’ for weighting purposes.
There was almost no weighting
required for region by age and gender.
Weighting was applied to bring education level in line with the general
population. Weights were applied to
increase the representation of those with some high school or a high school
diploma among all regions, and to decrease representation of those with a
University certificate or degree among all regions. Similarly, weights were applied to increase
representation of those with some high school or a high school diploma among
all age groups, and to decrease representation of those with a University
certificate or degree among all age groups.
Weighting was also applied to all regions to increase the representation
of those living in rural areas (under 1000 residents).
“Don’t know” and refused cases
were re-coded to groups with lowest counts. For education this was ‘Some high
school or less education’. For region, where postal code was available,
allocation was done based on postal code, and those without postal code were
re-coded to ‘Under 1000’. For Gender, males came up slightly lower to females
in their respective regions, hence “Don’t know” cases were allocated to males
for gender.
Base |
Unweighted |
Weighted |
||
2000 |
2000 |
|||
Effective
Base[6] |
2000 |
1394.2 |
||
|
Base |
% |
Base |
% |
Atlantic Canada Males |
|
|
||
18 to 34 |
16 |
1% |
16 |
1% |
|
|
|
||
35 to 54 |
21 |
1% |
21 |
1% |
55+ |
28 |
1% |
28 |
1% |
Atlantic Canada Females |
|
|
|
|
18 to 34 |
16 |
1% |
16 |
1% |
35 to 54 |
22 |
1% |
22 |
1% |
55+ |
31 |
2% |
31 |
2% |
Quebec Males |
|
|
|
|
18 to 34 |
61 |
3% |
61 |
3% |
35 to 54 |
76 |
4% |
76 |
4% |
55+ |
90 |
5% |
90 |
5% |
Quebec Females |
|
|
|
|
18 to 34 |
57 |
3% |
57 |
3% |
35 to 54 |
74 |
4% |
74 |
4% |
55+ |
99 |
5% |
99 |
5% |
Ontario Males |
|
|
|
|
18 to 34 |
118 |
6% |
116 |
6% |
35 to 54 |
125 |
6% |
125 |
6% |
55+ |
136 |
7% |
138 |
7% |
Ontario Females |
|
|
|
|
18 to 34 |
110 |
6% |
110 |
5% |
35 to 54 |
131 |
7% |
130 |
6% |
55+ |
152 |
8% |
156 |
8% |
Prairie Provinces Males |
|
|
|
|
18 to 34 |
58 |
3% |
57 |
3% |
35 to 54 |
59 |
3% |
62 |
3% |
55+ |
58 |
3% |
58 |
3% |
Prairie Provinces Females |
|
|
|
|
18 to 34 |
54 |
3% |
54 |
3% |
35 to 54 |
61 |
3% |
61 |
3% |
55+ |
62 |
3% |
62 |
3% |
BC & Territories Males |
|
|
|
|
18 to 34 |
41 |
2% |
41 |
2% |
35 to 54 |
46 |
2% |
45 |
2% |
55+ |
53 |
3% |
53 |
3% |
BC & Territories Females |
|
|
|
|
Females
18 to 34 |
38 |
2% |
39 |
2% |
Females 35 to 54 |
48 |
2% |
47 |
2% |
Females
55+ |
59 |
3% |
58 |
3% |
Base |
Unweighted |
Weighted |
||
2000 |
2000 |
|||
Effective
Base |
2000 |
1394.2 |
||
|
Base |
% |
Base |
% |
Atlantic Canada |
|
|
|
|
1000+ residents |
102 |
5% |
72 |
4% |
Under
1000 residents |
32 |
2% |
61 |
3% |
Quebec |
|
|
|
|
1000+ residents |
426 |
21% |
367 |
18% |
Under
1000 residents |
31 |
2% |
89 |
4% |
Ontario |
|
|
|
|
1000+ residents |
732 |
37% |
668 |
33% |
Under
1000 residents |
40 |
2% |
107 |
5% |
Prairie Provinces |
|
|
|
|
1000+ residents |
321 |
16% |
278 |
14% |
Under
1000 residents |
31 |
2% |
75 |
4% |
BC & Territories |
|
|
|
|
1000+ residents |
269 |
13% |
242 |
12% |
Under
1000 residents |
16 |
1% |
40 |
2% |
Base |
Unweighted |
Weighted |
||
2000 |
2000 |
|||
Effective
Base |
2000 |
1394.2 |
||
|
Base |
% |
Base |
% |
Atlantic Canada |
|
|
|
|
Some high school or less education / High
school diploma or equivalent / Apprenticeship or other trades |
42 |
2% |
74 |
4% |
College,
CEGEP, or other non-university certificate/diploma |
36 |
2% |
31 |
2% |
University certificate/diploma or degree |
56 |
3% |
28 |
1% |
Quebec |
|
|
|
|
Some high
school or less education / High school diploma or equivalent / Apprenticeship
or other trades |
159 |
8% |
259 |
13% |
College, CEGEP, or other non-university
certificate/diploma |
104 |
5% |
83 |
4% |
University
certificate/diploma or degree |
194 |
10% |
114 |
6% |
Ontario |
|
|
|
|
Some high school or less education / High
school diploma or equivalent / Apprenticeship or other trades |
204 |
10% |
378 |
19% |
College,
CEGEP, or other non-university certificate/diploma |
194 |
10% |
168 |
8% |
University certificate/diploma or degree
University certificate/diploma or degree |
374 |
19% |
228 |
11% |
Prairie Provinces |
|
|
|
|
134 |
7% |
193 |
10% |
|
College, CEGEP, or other non-university
certificate/diploma |
71 |
4% |
69 |
3% |
University
certificate/diploma or degree |
147 |
7% |
92 |
5% |
BC & Territories |
|
|
|
|
Some high school or less education / High
school diploma or equivalent / Apprenticeship or other trades |
85 |
4% |
147 |
7% |
College,
CEGEP, or other non-university certificate/diploma |
68 |
3% |
53 |
3% |
University certificate/diploma or degree |
132 |
7% |
82 |
4% |
Table
2.5.d. Dimension 4 – Education, Gender
and Age Unweighted
Base |
Unweighted |
Weighted |
||
2000 |
2000 |
|||
Effective
Base |
2000 |
1394.2 |
||
|
Base |
% |
Base |
% |
Males 18 to 34 |
|
|
|
|
Some high school or less education / High
school diploma or equivalent / Apprenticeship or other trades |
92 |
5% |
175 |
9% |
College,
CEGEP, or other non-university certificate/diploma |
55 |
3% |
49 |
2% |
University certificate/diploma or degree |
147 |
7% |
68 |
3% |
Males 35 to 54 |
|
|
|
|
Some high
school or less education / High school diploma or equivalent / Apprenticeship
or other trades) |
92 |
5% |
163 |
8% |
College, CEGEP, or other non-university
certificate/diploma |
71 |
4% |
67 |
3% |
University
certificate/diploma or degree |
164 |
8% |
99 |
5% |
Males 55+ |
|
|
|
|
Some high school or less education / High
school diploma or equivalent / Apprenticeship or other trades |
136 |
7% |
225 |
11% |
College,
CEGEP, or other non-university certificate/diploma |
76 |
4% |
56 |
3% |
University certificate/diploma or degree |
153 |
8% |
86 |
4% |
Females 18 to 34 |
|
|
|
|
Some high
school or less education / High school diploma or equivalent / Apprenticeship
or other trades |
83 |
4% |
123 |
6% |
College, CEGEP, or other non-university
certificate/diploma |
51 |
3% |
60 |
3% |
University
certificate/diploma or degree |
141 |
7% |
93 |
5% |
Females 35 to 54 |
|
|
|
|
Some high school or less education / High
school diploma or equivalent / Apprenticeship or other trades |
72 |
4% |
123 |
6% |
College,
CEGEP, or other non-university certificate/diploma |
101 |
5% |
90 |
4% |
University certificate/diploma or degree |
163 |
8% |
120 |
6% |
Females 55+ |
|
|
|
|
Some high
school or less education / High school diploma or equivalent / Apprenticeship
or other trades) |
149 |
7% |
242 |
12% |
College, CEGEP, or other non-university
certificate/diploma |
119 |
6% |
83 |
4% |
University
certificate/diploma or degree |
135 |
7% |
81 |
4% |
A total of 9658
invitations were sent to panelists, of which n=2,000 completed the survey. The overall completion rate achieved for
the online study was 67%. The following
table outlines the sample disposition and response rate as per the former Marketing
Research and Intelligence Association guidelines.
|
|
Total
Invitations Sent |
9658 |
Contacts |
2979 |
Completes |
2000 |
Break
Offs[7] |
367 |
Over
Quota[8] |
480 |
Non-Qualifiers[9] |
132 |
Completion
Rate |
67% |
Incidence
Rate |
80% |
A non-response bias analysis is the process
that results in the quantification of estimated nonresponse bias, and
identification of potential sources of nonresponse bias on estimates. To ensure a representative sample, this
research used completion quotas and as such a non-response bias analysis cannot
be undertaken.
As mentioned
previously, panel sample was used for this study. Panel surveys are considered a non-random
sample and as such margin of error does not apply.
The
discrete choice exercise was estimated using a Hierarchical Bayes Multinomial
Logit model. A detailed description of this approach can be found here https://datajobs.com/data-science-repo/Hierarchical-Bayes-%5BAllenby-and-Rossi%5D.pdf
The
model is called “hierarchical” because it consists of two models that are
jointly applied. One model is used for
within-respondent analysis, dealing with the internal heterogeneity in the
choice selections. The other model is
for a cross-respondent analysis and deals with external heterogeneity.
The
combination of the two models working simultaneously provided respondent-level
estimates of preference of the study’s attributes of visibility, health risk,
and annual household cost while factoring in each respondent’s age, region, gender,
household income and the health impact for yourself and your family.
The
Hierarchical Bayes Multinomial Logit model was estimated using Sawtooth
Software’s CBC Hierarchical Bayes Module v5.5.6. The model used an iterative Monte Carlo
Markov Chain approach to estimate the model for 200,000 iterations with the
first 100,000 iterations used as a burn-in to calibrate the process and the
last 100,000 iterations used to provide a robust estimate of the model. The final model estimated linear effects for visibility
and annual household cost and categorical effects for the two levels of health
risk. This model generated a robust estimate of the willingness to pay per
household for each one-unit decrease in the deciview (DV) scale for the entire
sample and for various subgroups of interest.
The linear
coefficients for visibility and annual household cost were normalized to
minimize scale bias of the estimates.
The visibility level values were calculated as – (DV – mean (DV))/10
where the mean DV was 21.
Visibility Shown |
Model Value |
|
33 DV |
10-15 km |
-1.2 |
29 DV |
20-25 km |
-0.8 |
25 DV |
30-35 km |
-0.4 |
21 DV |
45-50 km |
0 |
17 DV |
70-75 km |
0.4 |
13 DV |
105-110 km |
0.8 |
9 DV |
155-160 km |
1.2 |
The annual household cost level values were
calculated similarly as (Cost/Month – mean (Cost/Month))/10 so that both level
values are on the similar scales centered on zero.
Annual
Household Cost |
Model Value |
None |
-1.0 |
$30 ($2.50 per month) |
-0.75 |
$60 ($5.00 per month) |
-0.50 |
$90 ($7.50 per month) |
-0.25 |
$180 ($15.00 per month) |
0.5 |
$360 ($30.00 per month) |
2.0 |
eImproved(DV, Cost, Health
Risk) = eBaseline(DV,
Cost=$0, Health Risk)
and
removed the exponential from both sides of the equation and applied the actual
values and utilities, setting the improved DV value to be the baseline DV minus
one unit of DV:
DVBaseline*UtilityDV
+ Cost$0*UtilityAHC + UtilityHRLevelBaseline = (DVBaseline -1)*UtilityDV
+ CostImproved*UtilityAHC + UtilityHRLevelimproved
Then to
solve for Costimproved or WTP, we combined terms:
WTP =
((DVBaseline – DVBaseline + 1)*UtilityDV +
Cost$0*UtilityAHC + UtilityHRLevelBalseline -
UtilityHRLevelimproved) / UtilityAHC
And
when we applied the normalizations, the WTP calculation is:
WTP = (10*(-0.1*
UtilityDV - UtilityAHC + UtilityHRLevelBaseline
- UtilityHRLevelImproved)/ UtilityAHC) + 10
Definition |
|
DV |
Deciview |
AHC |
Annual Household Cost |
HR |
Health Risk |
Willingness to pay for 1 DV improvement in
visibility with no change to health risk
On average, Canadians are willing to pay $107.04
annually or $8.92 per month for an improvement of one DV to visibility. The median is $1.10 per month and the
standard deviation is $21.27 per month indicating a wide variability in the
amount that Canadians are willing to pay for 1 DV improvement in visibility. In total, 18 per cent[10] of Canadians are unwilling to pay any amount
for improvements to visibility which is contributing to the large variation. As
a reminder, protest responses were removed from the data – details of this
process can be found elsewhere in the report.
While there is wide variation in the amount Canadians are WTP, it should
be noted that that the annual value is in line with other Canadian research by
Haider et al, 2019[11] in which they report “respondents are willing to pay
a baseline amount of $92.52–$111.60 per year per household (Can$ 2002) for a
5%–20% increase in the visual range”.
‒ Younger Canadians
(18-34) are willing to pay significantly more for visibility improvements than
their older counterparts (35+).
Specifically, those 18-34 are willing to pay $165.96 annually ($13.83
per month) for 1 DV improvement to visibility while those who are 35-54 are
willing to pay $96.60 annually ($8.05 per month) and those 55+ are willing to
pay $70.92 annually ($5.91 per month);
‒ Households with
children are willing to pay more for visibility improvements compared to
households without children ($158.16 annually/$13.18
per month vs $85.08 annually/$7.09 per month);
‒ Households with
individuals whose health is impacted by air quality are willing to pay more
than those without (($143.64 annually/$11.97
per month vs $71.88 annually/$5.99 per month);
and
‒
Households where the existing
visibility is quite high generally (9 DV) are also more willing to pay than
those with generally lower visibility ($155.64
annually/$12.97 per month vs $88.68 – 107.04
annually/$7.39-8.92 per month)
|
Monthly Mean WTP for 1DV improvement |
Monthly Median WTP for 1DV improvement |
Standard Deviation for Monthly WTP for
1DV improvement |
Standard Error for Monthly WTP for 1DV
improvement |
Annualized WTP for 1 DV improvement
(monthly * 12) |
Total (A) |
8.92 |
1.10 |
21.27 |
0.49 |
107.04 |
AGE |
|
|
|
|
|
18-34 (B) |
13.83
CD |
1.60 |
29.74 |
1.27 |
165.96 CD |
35-54 (C) |
8.05
D |
0.96 |
18.89 |
0.76 |
96.6 D |
55+ (D) |
5.91 |
0.98 |
13.38 |
0.50 |
70.92 |
GENDER |
|
|
|
|
|
Male (E) |
8.03 |
0.95 |
19.87 |
0.65 |
96.36 |
Female (F) |
9.83 |
1.26 |
22.61 |
0.73 |
117.96 |
REGION |
|
|
|
|
|
Atlantic (G) |
4.98 |
1.14 |
18.21 |
1.62 |
59.76 |
Quebec (H) |
10.32 |
1.15 |
22.55 |
1.09 |
123.84 |
Ontario (I) |
8.25 |
0.99 |
20.51 |
0.76 |
99.00 |
Prairies (J) |
8.33 |
0.99 |
18.81 |
1.04 |
99.96 |
BC + Territories (K) |
11.07
G |
1.91 |
24.95 |
1.52 |
132.84 G |
EDUCATION |
|
|
|
|
|
High School or Less (L) |
8.40
|
1.08 |
19.73 |
0.93 |
100.80 |
College/CGEP(M) |
9.18 |
1.23 |
21.99 |
0.86 |
110.16 |
University+ (N) |
9.03 |
0.97 |
22.64 |
0.82 |
108.36 |
LANGUAGE |
|
|
|
|
|
English (O) |
8.58 |
1.11 |
20.96 |
0.54 |
102.96 |
French (P) |
11.41 |
1.09 |
25.13 |
1.35 |
136.92 |
Others (Q) |
11.50 |
1.48 |
22.66 |
3.17 |
138.00 |
CHILDREN |
|
|
|
|
|
Yes (R) |
13.18
S |
1.63 |
27.60 |
1.18 |
158.16 S |
No (S) |
7.09 |
0.97 |
17.81 |
0.49 |
85.08 |
COMMUNITY |
|
|
|
|
|
Urban (T) |
8.65 |
1.05 |
21.22 |
0.52 |
103.80 |
Rural (U) |
10.63 |
1.30 |
23.19 |
2.32 |
127.56 |
HEALTH IMPACT |
|
|
|
|
|
YES (V) |
11.97
Y |
1.56 |
25.79 |
0.85 |
143.64 Y |
Self (W) |
12.02
Y |
1.35 |
26.36 |
1.14 |
144.24 Y |
Family (X) |
12.49
Y |
1.63 |
26.39 |
1.09 |
149.88 Y |
No (Y) |
5.99 |
0.85 |
15.46 |
0.53 |
71.88 |
BASELINE |
|
|
|
|
|
18-33 DV (Z) |
7.39 |
0.85 |
21.85 |
1.03 |
88.68 |
17 DV (a) |
8.18 |
1.25 |
18.12 |
0.92 |
98.16 |
13 DV (b) |
8.92 |
1.24 |
20.83 |
0.92 |
107.04 |
9 DV (c) |
12.97
Zab |
1.63 |
25.96 |
1.37 |
155.64 Zab |
INCOME |
|
|
|
|
|
<$60k (d) |
8.21 |
1.02 |
19.67 |
0.71 |
98.52 |
$60k-$99k (e) |
10.26 |
1.08 |
24.49 |
1.04 |
123.12 |
$100k-$149k (f) |
7.85 |
1.14 |
20.59 |
1.21 |
94.20 |
$150k+ (g) |
8.67 |
1.71 |
18.08 |
1.61 |
104.04 |
Note: Letters denote statistically significant
difference within the column for each demographic group (p = 0.05). For
example, a D next to the result for 35-54 year olds under Monthly WTP for 1DV
improvement denotes this value is significantly greater than the value for 55
and older.
On
average, Canadians are willing to pay $581.76 annually or $48.48 per month for an improvement of one DV to visibility that includes a perceived
associated decrease in health risk from moderate to low. This research also examined WTP variation
based on differences in age, gender, region, education, language spoken most
often at home, community size, children in the household, health impact from air
quality and baseline or typical visibility for the respondent. Differences in WTP that includes an
associated decrease in health risk from moderate to low varied exist based on
age, gender, region, having children in the household and/or having a member of
the household (either family or self) whose health is impacted by air quality.
‒
Younger
Canadians (18-34) are willing to pay significantly more for visibility
improvements that included an associated decrease in health risk than their
older counterparts (35+). Specifically,
those 18-34 are willing to pay $897.72 annually
($74.81 per month) for 1 DV improvement to visibility while those who are 35-54
are willing to pay $498.36 annually ($41.53 per month) and those 55+ are willing to pay $412.56 annually ($34.38 per
month);
‒
Women are
willing to pay significantly more for visibility improvements that included an
associated decrease in health risk than men ($664.56
annually/$55.38 per month vs $499.56
annually/$41.63 per month);
‒
Households
in the Atlantic are willing to pay significantly less for visibility
improvements that included an associated decrease in health risk ($294.12 annually/$24.51 per month) compared to other
regions in Canada) ($525.36-682.08 annually/ $43.78-56.84
per month)
‒
Households
with children are willing to pay more for visibility improvements with
associated health improvements compared to households without children ($811.20 annually/$67.60 per month vs $473.88 annually/$39.49 per month); and
‒
Households
with individuals whose health is impacted by air quality are willing to pay
more than those without ($695.04 annually/$57.92
per month vs $435.12 annually/$36.26 per
month).
Note:
Letters denote statistically significant difference within the column for each demographic
group. For example, a D next to the result for 35-54 year olds under Monthly
WTP for 1DV improvement denotes this value is significantly greater than the
value for 55 and older.
In
order to provide more clarity around the WTP differences with improved health
risks, we reviewed the ratio of WTP on its own, compared to WTP with an
associated improvement to health risk.
Absolute
WTP values identify how much Canadians care about visibility and how much they
care about health. The ratio analysis
allows us to understand how much Canadians care about visibility compared with
health. Not unexpectedly, virtually all
Canadians care more about health than visibility. The analysis below helps us to identify “how
much more” various Canadians care about health over visibility.
Ratios
which are smallest provide some indication of which groups care the most about
visibility, relative to how much they care about their health. Based on the
ratio analysis, the following groups tend to care more about visibility:
·
Canadians
living in on the coast (east, west, or north) care more about visibility than
inland Canadians.
·
Rural
Canadians care more about visibility than urban Canadians, likely a function of
their baseline visibility being generally higher and supported by the finding
that Canadians living in areas with good visibility (9 DV or less) care more
about visibility than those living in poorer visibility (10 DV or more.
·
Canadians
with health conditions impacted by air quality care more about visibility than
those without, likely signalling the challenge in separating visibility and
health completely.
·
Canadians
in the lower income brackets care slightly more than those in higher income
brackets about visibility.
This analysis highlights a number of
interesting findings that generally show the gap between groups tends to be
narrower compared to when we look at absolute monetary WTP. It also highlights that a person’s baseline
impacts perceptions and desires for visibility and that perceptions of health
are difficult to remove in situations of poor visibility. More specifically, those who already have
good visibility are willing to pay more for improvements.
Note:
Letters denote statistically significant difference within the column for each
demographic group. For example, a D next to the result for 35-54 year olds under
Monthly WTP for 1DV improvement with no change to health risk denotes this
value is significantly greater than the value for 55 and older.
The survey was designed to identify when a
respondent was not willing to pay to improve visibility and determine whether
it was a true zero (unable to pay) or a protest response. Respondents that selected the baseline or “status quo”
option across all eight choice scenarios were asked for the reason for always
selecting the baseline option. The
intent of this question was to determine if the respondent truly was not
willing or able to pay any amount of money for a visibility improvement, or if
the responses were made in “protest” or because insufficient information was
given to make an informed decision. It
was determined that if the reason was given as “I would like to see improvements
but do not think I should pay for them”, “I object to the way the question was
asked”, or “I did not have enough information to base my decision” then the
respondent was removed from the sample and further analysis. In total, six per cent were considered
protest responses and were removed from the above analysis.
As stated
previously, the survey was designed to identify when a respondent was not
willing to pay to improve visibility and determine whether it was a true zero
(unable to pay) or a protest response. Among
those that consistently chose no changes to the baseline visibility option, the
majority (36%) stated this was because they would like to see improvements but
did not think they should have to pay for them.
Fewer said that improvements were necessary, but the options were too
expensive (21%), that the baseline option was acceptable and no improvements in
visibility were necessary (18%). A
minority answered that the baseline option was acceptable and no improvements
in health were necessary or that they did not have enough information to decide
(both 10%). In summary, six percent were
considered protest responses while eighteen percent were true zero responses.
Further,
among the few who said they did not have enough information to decide, the most
common information they said they would need included:
‒ More details, information, or facts (14%);
‒ More information on air quality (5%);
‒ Statistics for health and air quality (5%);
‒ How the cost to improve air quality was
calculated (5%); or
‒ Need to know how air quality could be improved
(5%).
When asked to select the image that best
represented the typical or average visibility range that they experienced in
the summer, respondents were most likely to select the two highest visibility
options. Few selected low visibility
options.
‒ 13 DV (105-110 km) (27%);
‒ 9DV (155-160 km) (22%);
‒ 17 DV (70-75 km visibility) (20%);
‒ 21 DV (45-50 km visibility) (10%);
‒ 25 DV (30-35 km visibility) (6%);
‒ 29 DV (20-25 km visibility) (4%);
‒ 33 DV (10-15 km visibility) (2%); and
‒ Don’t know (10%).[12]
As expected, those who live in a rural area are
more likely to report the highest visibility condition of 9 DV (155-160 km
visibility) than those who live in an urban area (43% vs. 18%). Those who live in the Atlantic provinces are
most likely to report an average visibility of 9 DV (155-160 km visibility)
(39%) while those in Ontario are least likely to report this visibility
compared to other provinces (19% vs. 23-26%).
Detailed tables are included under a separate
cover.
The
results of this research indicate, on average, Canadians are willing to pay
$107.04 per household annually or $8.92 per month for an improvement of one DV
to visibility. There is, however, large
variation in the amount that individual Canadians are WTP and noticeable differences
among different demographic groups. More
specifically, younger Canadians (18-34), households with children or with
individuals with health conditions impacted by air quality and those who
currently live in areas with high visibility are all willing to pay more than
their respective counterparts.
|
|
||||
|
||||
|
||||
|
||||
|
|
||
|
||
|
||
|
|
||||
|
||||
|
||||
|
||||
|
|
||||||
|
||||||
|
||||||
|
||||||
|
|
||||||||||||
|
||||||||||||
|
||||||||||||
|
||||||||||||
|
|
||
|
||
|
||
|
||
|
|
||||||||||||||||||||||||
|
||||||||||||||||||||||||
|
||||||||||||||||||||||||
|
||||||||||||||||||||||||
|
||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||||||||
|
|
||||||||||||
|
||||||||||||
|
||||||||||||
|
||||||||||||
|
|
|
|
||
|
||
|
|
||||
|
||||
|
||||
|
||||
|
||||
|
|
||||
|
||||
|
||||
|
||||
|
||||
|
|
||
|
||
|
|
||
|
||
|
||
|
|
|
||
|
||
|
|
||
|
||
|
||
|
||
|
|
||||||||||||||
|
||||||||||||||
|
||||||||||||||
|
||||||||||||||
|
||||||||||||||
|
||||||||||||||
|
|
||
|
||
|
||
|
||
|
|
|
|
|
||
|
||
|
|
||||
|
||||
|
||||
|
||||
|
|
||||||||||||||||||
|
||||||||||||||||||
|
||||||||||||||||||
|
||||||||||||||||||
|
|
||||||||
|
||||||||
|
||||||||
|
||||||||
|
||||||||
|
|
||||||||||||||||||||
|
||||||||||||||||||||
|
||||||||||||||||||||
|
||||||||||||||||||||
|
||||||||||||||||||||
|
|
||||||||||||||||
|
||||||||||||||||
|
||||||||||||||||
|
||||||||||||||||
|
|
||||||||||||||
|
||||||||||||||
|
||||||||||||||
|
||||||||||||||
|
|
||||||||||||||||
|
||||||||||||||||
|
||||||||||||||||
|
||||||||||||||||
|
||||||||||||||||
|
|
||||||||||||
|
||||||||||||
|
||||||||||||
|
||||||||||||
|
|
||||||||||||||
|
||||||||||||||
|
||||||||||||||
|
||||||||||||||
|
|
||||
|
||||
|
||||
|
||||
|
|
||
|
||
|
||
|
||
|
||
|
|
|
||
|
||
|
||
|
Appendix B: Air
Quality Images
Figure 4.2.d. 21 DV
(45-50 km)
Figure 4.2.e. 25 DV
(30-35 km)
Figure 4.2.f. 29 DV (20-25
km)
Figure 4.2.g. 33 DV
(10-15 km)
Programmer
Instructions: 100 design versions, 8 tasks per respondent, 2 choices per screen
1. Randomly assign
a respondent to one of the design versions, removing that version from the set
of available versions until all versions have been shown and then repeat full
list.
2. Randomly order
the tasks within version for each respondent
3. DO NOT randomize
the order of attributes within task, take them exactly as they are in the
design (see example task for question layout)
4. DO NOT randomize
the order of the options on the screen.
5. Make sure that
the grid fits on most screens without scrolling
Task Example
[1] Regulations Respecting
Reduction in the Release of Methane and Certain Volatile Organic Compounds
(Upstream Oil and Gas Sector) on April 26, 2018 (Canada Gazette, Part II, Vol.
152, Extra. SOR/DORS/2018-66 (pages 44 – 124))
[2] Multi-Sector Air Pollutants Regulations (MSAPR) on June 29, 2016 (Canada
Gazette Part II, Vol. 150, No.13. SOR/DORS/2016-151 (pages 1872 – 2175))
[3] Interagency
Monitoring of Protected Visual Environments (IMPROVE), 1993, Vol.2, No.1.
Deciview, A Standard Visibility Index
[4] Kantar data
indicates mail response rates in the 2-5% range and telephone response rates in
the 3-9% range
[5] http://www.airqualityontario.com/press/faq.php#:~:text=The%20Air%20Quality%20Health%20Index%20is%20a%20scale%20that%20lists,risk%20associated%20with%20air%20quality.&text=Scientists%20created%20the%20index%20by,on%20a%2010%20point%20scale.
[6]Statistically speaking,
a weighted sample generally has more sampling error than an unweighted sample
of the same size.
A
weighted sample's "effective base" size is the size of an unweighted
random sample that would have the same sampling error as the weighted sample.
[7] Respondents who
partially completed the survey
[8] Respondents who did not qualify for the survey
because they fell into demographic groups that already had the requisite number
of completes
[9] Respondents who did not qualify for the survey based
on exclusionary criteria (e.g., industry screen out for market research
employees)
[10] Protest responses removed
[11] Climate change, increasing forest fire incidence, and the value of
visibility: evidence from British Columbia, Canada in the Canadian Journal of
Forestry Research (July 2019)
[12]Note: Numbers do not add
to 100% due to rounding