Economic impacts of wind farms on Scottish tourism: report

6 The internet survey

6.1 Objective

The third major element of the study is an internet survey designed to explore the scenic value lost to the public when a wind farm is established. The only exogenous major factor that was thought might determine this value was the income of the individual respondent. However it was also believed that there was likely to be substantial variance between individuals. The approach was therefore to aim for maximum coverage at minimum cost ensuring in the design an allowance for income variance. Experience elsewhere and a promise of access to an extensive relevant email list suggested that an electronic survey would be the best approach

6.2 Contingent valuation

The contingent valuation method is the most direct valuation method and simply asks someone directly to state their maximum willingness to pay for a good or service. The method is well known and has been the subject of several books (Alberini, 2006; Bateman and Willis, 1999; Bjornstad and Kahn, 1996; Braden and Kolstad, 1991; Cummings et al, 1986, Mitchell and Carson, 1989). The technique was introduced in 1949 in an article by Ciriacy-Wantrup (Hanley et al, 2003 p. 3). The first application is provided by Davis (1963). In the early days of the technique, questions were open ended and were of the form 'What is the maximum you would be willing to pay for nice scenery while on holiday in Scotland?. Boyle and Bishop (1984) provide an early example of an attempt to value scenery.

The technique has come under significant scrutiny since its early days. Most of the concerns relate to whether people can give meaningful answers to open ended valuation questions and how their responses are influenced by survey design. These concerns were highlighted in the wake of the Exxon-Valdez oil spill in the US in 1989. A CV study was conducted to assess the environmental damage (including non-use values). Carson et al (2003) provide a review of the study. The study was heavily criticised (Diamond and Hausman, 1994) and as a result the National Oceanographic and Atmospheric Administration ( NOAA) commissioned a report on the technique. The report (Arrow et al, 1993) provides a review of the technique, the criticisms of the technique and what can be done to ensure robust results are obtained. Haab and McConnell (2003 pp. 20-22) summarise the key finding which relate to survey design.

One of the key findings was that the form of the question should be changed from open-ended to a referendum type question. With this form of question, the respondent is asked 'Would you be willing to pay £x to preserve Scotland's scenery in its current form?', where the value of x is different for different respondents. This approach is sometimes referred to as the dichotomous choice approach. It is believed that this style of question reduces bias in the results and significantly lowers the cognitive burden faced by respondents. Loomis (1988) discusses the differences in reliability between the open ended and the dichotomous choice methods. An example of this type can be seen in Bennett et al (2003) in the context of countryside access.

One of the problems with the method is that asking a dichotomous choice style question gives only one piece of information. For example, if someone is not willing to pay £30 for something, it is known only that their willingness to pay lies below £30. There is, however, a significant difference between £0.01 and £29.99. There is no way of knowing which is closer to the respondent's WTP. The open ended style question obtains (or at least aims to obtain) the precise figure. To combat this problem, Hanemann (1985) and Carson (1985) proposed asking a follow up question. If, for example, the respondent answered no to paying £30, they might be asked if they would pay £15. This would help to narrow down the range within which their true WTP lies. This approach is known as the double bounded dichotomous choice approach. Hanemann et al (1991) show this method to be more statistically efficient. The method is not without problems though (Carson et al., 1992; Cameron and Quiggin, 1994; McFadden and Leonard, 1993; Kanninen, 1995). The main problem relates to the behaviour of the respondent. When asked the first question the respondent gives an 'honest' answer. When asked the second question, the mindset of the respondent changes to a 'bidding game' mindset. This renders the second answer inconsistent with the first (Barreiro, 2005).

It is often felt that hypothetical answers to hypothetical questions cannot provide robust results. Consequently most of the studies try to include an aspect which makes the respondent believe they will actually be required to pay the amount requested. Other approaches to assessing the reliability of WTP estimates have compared the stated preference results to revealed preference results (Brookshire et al, 1982; Carson et al., 1996). Such studies have shown that similar results are obtained using both methods.

Despite the issues surrounding the use of CV studies, and the considerable expense of dealing with these problems, the technique has been very popular. This is partly due to the fact that it can be used to measure the value of anything. Countless examples are available: Fix and Manfredo (2005) and the value of wildlife; Alonso (2002) and the value of accessible housing; Bateman and Langford (1997) and the value of national parks to non-users; Yoo at al (2006) and the cost of Spam email; Treiman and Gartner (2006) and the value of forests; Green and Tunstall (1991) and the value river water quality improvements and even the value of silence (Barreiroet al, 2005).

6.3 Design

Contingent Valuation Methods are normally based on face to face interviews. A few have attempted self response mail questionnaires but as far as can be ascertained none have used the internet approach. As discussed above in order to elicit sensible WTP results it is important that the respondent understands exactly what is being paid for and exactly how they will be paying for it. One of the advantages of conducting a face to face survey is that the interviewer can explain to the respondent what is happening. Because this survey is being administered online, a scenario which was easy to understand was needed.

It was decided that the respondents should be asked to choose between two rooms at a hotel. One room would have no landscape view (a view of the car park) while the other would have a view of the landscape. As an initial test of the concept, it was decided that respondents would be asked to perform this task 20 times. Each time, one alternative would be the car park view and the other would be a different scene each time. One- third of these scenes would be plain views of hills or water etc., while another third would have the same scenes but with some wind turbines, pylons, telegraph poles or deforestation added. The final third would be the same scenes but with even more of these built features present. The basic idea was that this approach could be used to measure how sensitive people are to seeing any alteration to the environment and then to measure how sensitive they are to the magnitude of the change. It was not clear at this stage if people would be able to understand what was demanded of them and, indeed, if the results generated would make any sense.

The basic survey design was as follows. Firstly respondents were presented with a story about booking accommodation; a standard double room at a 3 star hotel. They were then asked what their maximum willingness to pay for such a room would be. On the next 20 screens they were presented with the car park view put next to the view on offer. They were then asked their maximum willingness to pay to upgrade to the view on offer. Of course, they could choose to pay nothing to upgrade i.e. they would not move rooms or they could choose not to stay in the room with the car park view. The inclusion of this opt out option is important for reasons already discussed.

There were two main reasons for including things other than wind turbines in the photographs. The first was a genuine interest in how tourists respond to different kinds of features on the landscape. The second was to mask the fact that the survey was about wind farms. It was feared that anti and pro wind farm groups might try to manipulate the results of the research if they found out its main purpose.

In addition to these key questions, standard profiling questions were asked in order to test whether WTP figures were dependent on demographic differences and to ensure that the sample who answered the questionnaire was representative of Scottish tourists as a whole. One of the questions asks the respondent what their typical daily expenditure is when on a holiday in Scotland. This is important to make allowance for income differentials when using the willingness to pay to assess the likely economic impact. It also serves another function. One of the key elements in designing a CV study according to Arrow et al (1993 pp. 59-60) is to remind respondents of their budget constraints and alternative uses of the money which they state they would be willing to pay for whatever is on offer (i.e. an improved view). Asking expenditure at the start of the survey helps to remind people how much they would spend per day and therefore what percentage of this they would be spending if they paid extra for a room upgrade.

6.4 Survey construction

For the survey, photographs were needed of various types of scenes. Most of the photographs used were taken over the course of a week and some use was made of photographs already available. Pictures were taken of Braes of Doune Wind farm near Stirling and Earlsburn wind farm in the Campsies, also near Stirling. Other features represented in the pictures were deforestation, pylons and telegraph poles.

The next stage was to modify the core scenes to be clear of their key features (turbines, pylons etc) and to extend their features. This idea is not new and has been used in other CV studies (e.g. Brandolini, 2004). The software chosen to do make the modifications was Adobe Photoshop 7.0 (2002). This is the market leader in the area and has been used in other valuation studies for the same purpose (e.g. Alvarez-Farizo & Hanley, 2002).

SNAP Surveys (2007) was the software initially used to construct the questionnaire. The software makes it simple to ask the most straightforward kind of questions e.g. entering a number for age, or making a multiple choice selection for accommodation type (hotel, self-catering etc). It was decided that rather than give open ended WTP questions that respondents should be able to choose from a drop down list of price ranges. This both speeds up completion and goes some way to presenting the valuation as a choice, as advocated by Savage.

Construction of the photographic section of the survey was more difficult. After some experimentation it was found that externally matching the size and detail of the photographs to the package was essential (as opposed to merely importing the photograph) to cope with different screen sizes and resolutions. No information or detail is lost and reduces the length of time the survey takes to download.

6.5 The pilots

The survey was shown to some Glasgow Caledonian University colleagues before proceeding to a full scale pilot. Around 10 people completed the survey and found that it worked well and that they were able to understand it. It took around 5 minutes to complete and all those who took it reported that it was enjoyable.

For the full pilot, the survey was uploaded to the university's server and the link was sent to the staff email list. Although this was during a holiday period and many staff were not available, over 100 responses were obtained within a day as well as some comments on the survey. Respondents were asked not only to complete the survey but to email comments on design. The results gratifyingly appeared consistent with expectations and the comments largely both positive and helpful.

6.6 Randomizing question order

One key problem identified in the early stages was anchoring; that values set by the respondent in early questions tended to affect the values set in later questions. A typical thought pattern would be "I gave that a value of £15 and I like this one better". An excellent discussion of anchoring is presented in Green et al (1998).

The basic design had been sent to an external expert for comment and he was concerned both about the initial length of the survey and also suggested that it would be better if the order in which the scenes were presented was random.

The possibility of randomising the order of the questions was investigated and it was found that the SNAP "Survey Plus" toolbox contained a Randomize tool. One of the key features of the tool is that it allows portions of the survey to be randomised, and not just the survey as a whole. This was important since the profiling questions were required to be displayed first and the screen thanking the respondents for their participation had to be displayed last. Despite initial problems, which required a patch from the company's website, the eventual design proved a perfect solution to an important problem.

6.7 Publication and distribution

For the internet SNAP generates a set of HTML files. These were then uploaded to the public server at Glasgow Caledonian University which allowed them to be accessed from any location by clicking on the URLwww.gcal.ac.uk/econsurv/land

This process proved completely trouble free.

The next stage was to circulate the survey to a set of respondents who would be willing to click on the URL and undertake the survey. Ideally we required a very large email list of individuals likely to be interested in Scottish scenery. VisitScotland, the national tourist organisation, would have been the ideal vehicle through which to access such a list. Unfortunately data protection arrangements with their list members prevented any communication about research that had not been specifically commissioned by the organisation.

Despite a search for a single large alternative, none could be found. One alternative which was progressed was the equivalent of a snowball sample. Email lists of the consulting team were used and key contacts on the email list were then asked to circulate their personal lists with the URL. In addition the Operational Research Society, the Economics Teaching Exchange and the Countryside Network agreed to circulate their members asking them to circulate the URL.

Whilst it may be argued that the population surveyed is likely to be more random than that from a single list distribution there was considerable concern that a strong bias may emerge. As an example one of the authors is keen on outdoor activities and the email list in this case is dominated by members of the local canoe club and of the Scout Association. Any bias in this list towards placing a high value on scenery is likely to snowball via the contacts of the initial contacts. In addition there was a worry that the lack of control made the survey vulnerable to concerted action by those either committed or opposed to Wind farm developments.

One alternative that emerged late in the scheme was the use of panels developed by commercial companies. Because of technical difficulties this eventually involved a rescripting of the survey for different software Net- MR and distribution via the GMI (Global Market Insight) system. In fact two surveys were constructed. The first, designed for a UK general panel was identical to that produced using SNAP and shown in Appendix II. The second was designed for the US panel who had been screened to include only those who had visited Scotland or would do so in the near future. The major differences were the omission of the home country and the use of dollars rather than pounds sterling. Inclusion of other countries was possible but thought to be too expensive for any gain in information.

The size of the commercial panels results in invitations to participate only going to a fraction determined by the target set. For the UK this target was 600 responses with an age and gender distribution reflecting that of UK tourists in Scotland. For the US the target was simply 100 who had been or were likely to go to Scotland in the near future. Because potential respondents will not be able to complete the survey once the target has been met, a conventional response rate cannot be calculated. Response rates on internet surveys are known to be low and, even with incentives, in the UK and US are unlikely to exceed 15%.

6.8 Processing and output

One of the major advantages to electronic surveys is that data processing is automatic. SNAP for example identifies responses from the email subject title and then simply records and processes the message content. Whilst the software incorporates statistical software which is particularly strong for data presentation, it also provides a facility to export the data in SPSS (.sav) format.

Net- MR works in a similar fashion and eventually produces a similar SPSS file of results. These files then had to be processed to obtain the percentage change in the willingness to pay. Firstly respondents were required to indicate what they would be willing to pay in terms of an interval e.g. £35-£50. The coded interval was recoded as the value of the mid-point of the range e.g. code 7 (£35-£50) would be recoded as £42.50.

Table 6-1gives a brief description of the pictures shown by each question and the derived variables.

Table 6-1 Variable Descriptions and Derived Variables

	Category	Description
Q10	Basic Price with View of	Car Park
Q11	Extra For View of	Braes of Doune without wind turbines
Q12		Braes of Doune wind farm (current)
Q13		Braes of Doune wind farm Extended
Q14		Bay near Thurso without wind turbines
Q15		Bay near Thurso with wind farm (planned)
Q16		Bay near Thurso with extended wind farm
Q17		Waterfall without wind turbines
Q18		Waterfall with wind turbines
Q19		Falkirk scene with No Grid Lines
Q20		Falkirk scene with 1 Grid Line
Q21		Falkirk scene with 2 Grid Lines
Q22		River Spey without Poles
Q23		River Spey with telegraph Poles
V1=Q12-Q11	Loss of Value from	Initial Build of Braes of Doune
V2=Q13-Q12		Extension at Braes of Doune (additional loss)
V3=Q15-Q14		Initial Build at Thurso
V4=Q16-Q15		Extension at Thurso (additional loss)
V5=Q18-Q17		Wind Turbine at Waterfall
V6=Q20-Q19		Falkirk scene - 1 Grid Line
V7=Q21-Q20		Falkirk scene - Extra Grid Line (additional loss)
V8=Q23-Q22		Telegraph Poles on Spey

The loss of value as a percentage of the room price (V/Q10) for each individual was then calculated and the mean percentage loss of value for the sample followed.

In the following sections we present the basic results for the surveyed populations and analyse how these differ.

6.9 UK Results

6.9.1 The Respondents

Age, Gender and Home

Table 6-2, Table 6-3 and Table 6-4 show the gender, age and home of the 606 respondents in the UK Survey.

Table 6-2 Distribution of Respondents by Gender

	Number	Percent
Male	303	50.0
Female	303	50.0
Total	606	100.0

Table 6-3 Distribution of Respondents by Age Group

	Number	Percent
16 - 25	72	11.9
26 - 45	255	42.1
46 - 65	210	34.7
Over 65	69	11.4
Total	606	100.0

Table 6-4 Distribution of Respondents by Residence

	Yes		No		Total
Highlands of Scotland	8	1.8%	0	0.0%	8	1.3%
Central Scotland	38	8.7%	0	0.0%	38	6.3%
Rest of Scotland	12	2.7%	0	0.0%	12	2.0%
North of England	109	24.9%	14	8.3%	123	20.3%
Midlands of England	79	18.1%	47	27.8%	126	20.8%
Southern England	163	37.3%	92	54.4%	255	42.1%
Ireland	8	1.8%	0	0.0%	8	1.3%
Mainland Europe	2	0.5%	1	0.6%	3	0.5%
Rest of World	1	0.2%	0	0.0%	1	0.2%
Wales	17	3.9%	15	8.9%	32	5.3%
TOTAL	437	100.0%	169	100.0%	606	100.0%
Percentage Visited	72.1%		27.9%		100.0%

The sample is broadly representative of the UK population with a significant number in the over 65 category. A significant majority (72%) have visited Scotland at some time. The majority of those who have not are, not surprisingly, located in the South of the UK. Tourist numbers are far more heavily weighted towards Scotland because of multiple repeat visits. Consequently it would have been inappropriate to sample on the basis of home locations of tourists.

With the information available it seems reasonable to conclude that we have a representative sample to identify the value that current and potential tourists from the UK would place on changes in the Scottish landscape.

Accommodation and Activities

Table 6-5 provides details of the accommodation used. It is believed that the majority of the other category is in the homes of "Friends and Family".

Table 6-5 Main type of accommodation used by sample

	Number	Percent
Hotel	203	46.3
Bed and Breakfast, Hostel	87	19.9
Hired Caravan	5	1.1
Caravan, Campervan, Tent	36	8.2
Self Catering	45	10.3
Other	62	14.2
Total	438	100.0

The primary reason for the trip is shown in Table 6-6.

Table 6-6 Principle Reason for Visit

	Number	Percent
To see Scotland	209	47.7
To see friends and relatives	105	24.0
To go shopping	13	3.0
Business trip	27	6.2
To see Scotland as an extension of a business trip	4	0.9
Personal business (appointment with doctor, dentist, solicitor)	4	0.9
To undertake a cultural activity (theatre visit, concert etc)	23	5.3
To participate in a sporting or outdoor activity	21	4.8
To watch a sporting activity	5	1.1
Other	27	6.2
Total	438	100.0

The sample has fewer trips where the principle reason was business than might be expected from the VisitScotland data. However many business trips are likely to be repeated within a year resulting in higher numbers of visits on business than recorded in this sample. In addition it is quite possible that those visiting on business also visit for holiday reasons as recorded here.

On the basis of the sample and with the assumption discussed in Chapter 2, we would expect those engaged on a Holiday Trip, Seeing Friends and Relatives and Participating in a sporting or outdoor activity (76.5%) to have a particular interest in landscape.

6.9.2 The Willingness to Pay for Views

Value of Scenery

The value placed on a scene is a function not only of the landscape but of the weather in which it is viewed. To identify the impact of structures, the report concentrates on the change in value between at same scene. However it is of interest to examine the "values" of the untouched scenes as in Table 6-7.

Table 6-7 The Value of Scenery

	Value of Scene
Braes of Doune	£22.71
River Scene (Spey)	£21.98
Rural near Falkirk	£15.87
Waterfall	£17.41
Bay near Thurso	£24.29
Average	£20.45

This table shows clearly that a good view is extremely valuable and important to a hotel, averaging £20 per room. The implication in terms of planning policy is obvious.

The average price for the room without the view was £40.96, suggesting that a good view could generate a 50% mark-up.

Value of Scenic Change by Location and Type

Table 6-8 provides estimates of the loss of scenic value to the average tourist when different types of developments occur in different locations. The most disliked was the pylon which caused an almost 30% drop in the value of the room, which, under the assumptions discussed earlier, will lead to a 30% fall in expenditure for the affected rooms.

Table 6-8 Loss of Value by Location and Type

	Loss £	Loss %
Loss for Braes	£6.56	18.8%
Additional Loss for Braes Extension	£1.54	6.5%
Total Loss for Extended Braes	£8.10	25.7%
Loss for Thurso	£6.17	16.6%
Additional Loss for Thurso Extension	£0.55	3.9%
Total Loss for Extended Thurso	£6.72	20.6%
Loss for Waterfall Development	£7.97	18.7%
Loss for Grid Line	£9.54	24.6%
Additional Loss for Second Grid Line	£1.22	4.5%
Total Loss for Both Grid Lines	£10.76	29.1%
Loss for Telegraph Poles	£4.58	11.7%
Basic Wind Farm Average Loss	£6.90	18.0%
Extended Wind Farm Average Loss	£7.41	23.2%

The loss for the wind farms varies from £6.17 (16.6%) for the basic Thurso development, to £8.10 (25.7%) for an extended Braes of Doon. A surprising and important result is the diminishing marginal loss associated with increasing size. It appears that once there has been an intrusion into the scenery then the effect of expanding the size is relatively small. This in turn suggests concentrating wind farm development would ceteris paribus be preferable to dispersion.

This finding essentially contradicts the finding of the intercept study and throws light on a number of anomalies in research in this area. Respondents to the internet survey are simply faced with a scene against the car park, there is no direct comparison between extended and basic farm. If we take the example of Thurso, individuals object to the wind farm whatever the size. In the internet study the doubling of the size is difficult to reference, particularly as the order of appearance is random. On the other hand if we ask the same respondents about the impact of increasing the size the response is likely to be strongly negative. Indeed we suggest that if the extended view had been referenced to the basic level rather than the car park we would have found a far more significant loss of value.

We find the same sort of problem later where actual reactions to existing wind farms are significantly smaller than the stated reactions in the internet survey. There is clearly a difference between actual and stated reactions and actual and stated values, with the actual being substantially lower than the stated.

It is a matter of conjecture why some developments appear more objectionable than others. The waterfall picture is undoubtedly the least "natural" and the foreground/weather on the Thurso photos the most pleasant with the turbines furthest away. To compute an average wind turbine loss, the loss for the Braes has been added to the loss for the waterfall and the loss for Thurso. This loss is now discussed in relation to the characteristic of the individual respondents.

Loss of value by age, gender and home location

Table 6-9 shows the mean loss of value by gender. Although females appear to place a higher value on the scenery the difference is not significant even at the 10% level because of the high variances and associated high standard errors of the means.

Table 6-9 Loss of Value by Gender

	Loss £	Loss %
Male	£6.94	15.6%
Female	£7.23	24.1%
Total	£7.08	19.7%

Table 6-10 shows the loss of value from wind turbines by age class. What is striking is the much lower value placed by the young on the scenery. This may reflect more familiarity with wind farms, a better capacity to adjust or, possibly, a lower income. The difference in absolute values is highly significant (t=3.116) but is only significant at the 10% for the percentage figures.

Table 6-10 Loss of Value and Age of Respondent

	Loss £	Loss%
16 - 25	£2.86	10.2%
26 - 45	£7.97	21.0%
46 - 65	£7.66	24.1%
Over 65	£6.47	11.7%
Total	£7.08	19.7%

For the elderly a major difference is the higher price for basic accommodation. Despite the apparent differences, unless one excludes the young, the elderly are not significantly different for the group as a whole.

The impact of location on valuation of scenery is shown in Table 6-11. Contrary to what might have been hypothesised the highest values seem to be associated with predominantly rural areas in the Highlands and Ireland. Once again wide variances and small numbers make it impossible to confirm this observation statistically.

Table 6-11 Loss of value by home region

	Loss £	Loss %
Highlands of Scotland	£12.22	38.0%
Central Scotland	£7.04	18.0%
Rest of Scotland	£6.19	20.1%
North of England	£7.80	22.9%
Midlands of England	£6.64	15.5%
Southern England	£6.84	20.1%
Ireland	£12.59	34.4%
Mainland Europe	£3.61	11.9%
Rest of World	£5.42	13.4%
Wales	£6.14	18.7%
Total	£7.08	19.7%

Expenditure, Income and Value

Figure 6-1 shows the distribution of the prices respondents were expecting to pay for the "standard room". It was expected that this might reflect income inequalities but it was found that there was little correlation with the typical spend reported as shown in Table 6-12.

Figure 6-1 Distribution of Room Prices

Table 6-12 Price of Room v Daily Expenditure

Daily Expenditure	Price of Room
More than £500	£43.50
£250-£500	£43.10
£150-£249	£43.10
£0-£149	£40.51
Total	£41.80

If the assumption is made that those with high daily expenditures tend to have high incomes and that those with high incomes tend to place a greater value on scenery then it follows that the percentage of the value of a room attributable to scenery should be more equal than the absolute. Table 6-13 shows that whilst there is some evidence of rising values with rising expenditure the variance of the percentage change is equally large. In fact in neither case are the differences statistically significant, and thus we conclude that there is little significant difference in valuations by expenditure (income).

Table 6-13 Relationship between Value of Scenery and Daily Expenditure

	Lost Value £	Lost Value%
More than £500	8.4	16.1%
£250-£500	7.6	18.0%
£150-£249	8.6	20.6%
£0-£149	7.0	20.5%
All	7.5	19.8%

Value and Visits to Scotland

One hypothesis that has been suggested is that visitors to Scotland tend to value landscape more than the average tourist. Table 6-14 shows the relative values.

Table 6-14 Value of Scenery and Visits to Scotland

	Visited Scotland	Mean
Loss £	Yes	£7.54
	No	£5.91
Loss %	Yes	19.8%
	No	19.6%

Although the absolute values appear to confirm the hypothesis, once again the difference is not statistically significant. In terms of percentage loss there is clearly no distinction.

Scenic Value, Accommodation and Activity

The relationship between value and accommodation in Table 6-15 shows similar consistency.

Table 6-15 Value and Accommodation

	Loss £	Loss %
Hotel	8.75	26.2%
Bed and Breakfast, Hostel	6.01	14.1%
Hired Caravan	3.58	16.3%
Caravan, Campervan, Tent	7.38	17.3%
Self Catering	6.16	19.4%
Other	7.03	8.5%
Total	7.53	19.8%

The cheapest hired accommodation (hired caravan) has the lowest absolute rate but as a percentage of the price paid is in line with other forms. Hoteliers tend to have most to lose from scenic deprivation which probably reflects the higher age ranges attracted.

Table 6-16 Value and Trip Purpose

	Loss	Loss%
To see Scotland	£7.34	18.6%
To see friends and relatives	£7.88	19.0%
Shopping and Business	£8.87	33.9%
Other	£6.78	14.9%

Table 6-16 examines the relationship between value and trip purpose. Once again there are no significant differences.

The range of values for individuals

The analysis so far has suggested that the only group that places significantly different values on the loss of landscape are the young. In part, this is because real differences are swamped by differences between individuals. Most individuals appear to prefer a landscape without a wind farm but there is also a substantial proportion that does not care (and a few who positively like wind farms). Table 6-17 shows this distribution.

Table 6-17 Distribution of values placed on changes by individuals

Percentiles	Braes Current	Braes Extended	Additional Value from Extension Braes	Single Grid Line	Double Grid Line	Additional Value from Extra Pylon	Waterfall	Spey	Thurso Current	Thurso Extended	Additional Value from Extension Thurso
10	-£20.00	-£25.00	-£10.23	-£26.25	-£27.00	-£8.50	-£26.48	-£20.00	-£20.00	-£21.00	-£8.50
20	-£12.50	-£12.50	-£3.75	-£18.00	-£20.00	-£4.00	-£14.00	-£10.00	-£12.50	-£12.50	-£3.75
30	-£8.50	-£9.25	-£1.25	-£12.50	-£12.50	-£0.96	-£9.93	-£6.25	-£9.25	-£9.25	£0.00
40	-£6.25	-£7.00	£0.00	-£8.50	-£10.00	£0.00	-£7.00	-£2.94	-£6.02	-£6.25	£0.00
50	-£3.75	-£5.50	£0.00	-£6.25	-£8.50	£0.00	-£3.75	£0.00	-£2.50	-£3.75	£0.00
60	-£0.96	-£2.50	£0.00	-£4.00	-£6.25	£0.00	-£1.50	£0.00	£0.00	£0.00	£0.00
70	£0.00	£0.00	£0.00	-£2.50	-£3.75	£0.00	£0.00	£0.00	£0.00	£0.00	£0.00
80	£0.00	£0.00	£0.00	£0.00	-£0.75	£0.00	£0.00	£0.00	£0.00	£0.00	£0.00
90	£0.00	£0.00	£4.83	£0.00	£0.00	£0.29	£0.00	£2.50	£0.00	£0.00	£7.45
Negative	61.70%	68.20%	32.20%	78.90%	81.70%	62.50%	62.50%	47.00%	56.10%	59.10%	22.70%
Neutral	29.20%	22.40%	49.70%	15.80%	14.00%	29.00%	29.00%	39.80%	34.20%	32.00%	52.60%
Positive	9.10%	9.40%	18.10%	5.30%	4.30%	8.50%	8.50%	13.20%	9.70%	8.90%	24.70%

This confirms quite clearly the relative indifference to size of Wind farms (Braes Extension and Thurso Extension) and the general dislike of grid lines and pylons (Double Grid Line). As far as Wind farms are concerned the pattern seems to be that the averages are Negative 63.3%, Neutral 27.8% and Positive 8.9%.

Summary on Value Estimates

There is a wide variance in values placed by individuals on the scenery that almost completely swamps any group characteristics. Given these findings it seems appropriate to treat the respondents as a homogeneous group and to utilize means for the whole group when assessing potential losses of value and consequential economic impact.

6.9.3 Perceptions and Reactions

The final section of the study sought tourist perceptions of the number and spread of wind farms in Scotland. There are two quite surprising findings shown in Table 6-18. Firstly there is the (incorrect) belief that turbines are as prevalent in scenic areas as in non-scenic areas.

Table 6-18 Prevalence of Wind Farms

	Non-Scenic		Scenic
	Frequency	Percent	Frequency	Percent
Very likely	36	5.9	33	5.4
Quite likely	161	26.6	148	24.4
Likely	186	30.7	198	32.7
Not very likely	213	35.1	204	33.7
Not at all likely	10	1.7	23	3.8
Total	606	100.0	606	100.0

Secondly there appear to be an exaggerated belief that one is currently likely to see a wind farm on a 2 hour journey. As discussed in chapter 5, routes to the west of the country are (M74 and A82/3) are still clear and planning permission has largely prevented developments in scenic areas. This situation may not last.

The final table summarises the responses to the question "If the number of wind farms in non scenic areas increases, what will be your likely response?"

Table 6-19 Possible Reaction to increase in number of wind farms

	Frequency	Percent
Go to see them	114	18.8
No response	374	61.7
Avoid the areas	108	17.8
Avoid Scotland	10	1.7
Total	606	100.0

On the positive side there is clearly a latent demand for a visit to a wind farm as part of the tourist experience. On the negative side these figures are very similar to those found in the much criticised System3 (2002) study and which have led to so much worry. They are noticeably different from the results of the "on the ground" intercept study and in reality these figures may well be exaggerated. One test is the difference in perception between those who have visited Scotland and those who have not.

Table 6-20 Difference in perception between visitors and non-visitors of likelihood of seeing Wind farm

	Non Scenic			Scenic
	Visited	Not Visited	Total	Visited	Not Visited	Total
Very likely	4.8%	8.9%	5.9%	5.0%	6.5%	5.4%
Quite likely	30.4%	16.6%	26.6%	23.8%	26.0%	24.4%
Likely	27.7%	38.5%	30.7%	30.2%	39.1%	32.7%
Not very likely	35.7%	33.7%	35.1%	37.5%	23.7%	33.7%
Not at all likely	1.4%	2.4%	1.7%	3.4%	4.7%	3.8%
	100%	100%	100%	100%	100%	100%

Those who have visited Scotland can clearly distinguish the policy of protecting scenic areas. Perhaps there is an argument for identifying the many scenic areas more clearly for visitors and the caution associated with their classification.

Table 6-21 Differences in Reaction between visitors and non visitors

	Visited	Not Visited	Total
Go to see them	17.8%	21.3%	18.8%
No response	61.6%	62.1%	61.7%
Avoid the areas	19.2%	14.2%	17.8%
Avoid Scotland	1.4%	2.4%	1.7%
	100.0%	100.0%	100.0%

As Table 6-21 shows the only difference (not significant) in reaction between those who have and those who have not visited Scotland, is avoiding the country rather than the areas. This probably reflects lack of information about the size and its variability of Scotland, but may also indicate a problem in the future about attracting new visitors.

6.10 US results

6.10.1 Respondents

A title of the project was circulated to the US panel, which for the purposes of this study could be regarded as random, and an invitation issued to respond to the survey. Results were obtained from the first 100 who have visited Scotland or plan to do so within 5 years. The number screened out was a surprisingly low 85, almost 55% of the initial sample had been or intended to visit Scotland.

Table 6-22 and Table 6-23 show the age and gender of this sample. It is suspected that the retired tourist is possibly under-represented but this does not have any significant impact (see section 6.10.3)

Table 6-22 Gender of US Respondents

	Frequency	Percent
Male	53	51.5
Female	50	48.5
Total	103	100.0

Table 6-23 Age of US Respondents

	Frequency	Percent
16 - 25	12	11.7
26 - 45	48	46.6
46 - 65	39	37.9
Over 65	4	3.9
Total	103	100.0

A significant majority of the sample used hotels, with the balance being taken up with cheaper indoor accommodation.

Table 6-24 Accommodation used by US Respondents

	Frequency	Percent
Hotel	70	68.0
Bed and Breakfast, Hostel	30	29.1
Caravan, Campervan, Tent	1	1.0
Self Catering	1	1.0
Other	1	1.0
Total	103	100.0

Table 6-25 Primary Purpose of US Tourists

	Frequency	Percent
To see Scotland	68	66.0
To see friends and relatives	7	6.8
To go shopping	1	1.0
To see Scotland as an extension of a business trip	5	4.9
To undertake a cultural activity (theatre visit, concert,	5	4.9
To participate in a sporting or outdoor activity	3	2.9
Other	14	13.6
Total	103	100.0

Table 6-25 shows that the vast majority are simple tourists with the next largest item being for "other" reasons. If we discount this group then it appears that 76.3% of the group would be directly affected by the scenery, remarkably close to the 76.5% of the UK sample.

6.10.2 The Willingness of US Tourists to Pay for Views

Value of Scenery

Table 6-26 compares the value placed on the scenes by US and UK tourists. The most striking features are the willingness of the US tourist to pay more for the view than the UK tourist and the similarity of the rankings of the scenes.

Table 6-26 Comparison of the value of specific scenes to US and UK tourists

	US	Rank	UK	Rank
Braes of Doune	£26.02	3	£22.71	2
Spey	£29.18	2	£21.98	3
Rural	£21.16	5	£15.87	5
Waterfall	£23.43	4	£17.41	4
Bay near Thurso	£30.45	1	£24.29	1
Average	£26.05		£20.45

The mean price for the room with the view of the car park only was £40.81, compared to £40.96 for the UK sample.

Value of Scenic Change by Location and Type

Table 6-27 shows the loss in value to US tourists compared to the loss for UK Tourists. Whilst they are of the same magnitude it is noticeable that the US tourist experiences less loss of value with wind farms than the UK tourist , despite placing a greater value on the scene. The one glaring exception is the impact of grid lines which are even more offensive to the US eye.

Table 6-27 Loss of value from developments for US and UK Tourists

	US		UK
	Loss £	Loss %	Loss £	Loss %
Loss for Braes	£4.66	6.2%	£6.56	18.8%
Additional Loss for Braes Extension	£2.61	9.3%	£1.54	6.5%
Total Loss for Extended Braes	£7.27	15.7%	£8.10	25.7%
Loss for Thurso	£6.08	7.3%	£6.17	16.6%
Additional Loss for Thurso Extension	-£0.07	2.7%	£0.55	3.9%
Total Loss for Extended Thurso	£6.02	10.0%	£6.72	20.6%
Loss for Waterfall Development	£5.95	12.7%	£7.97	18.7%
Loss for Grid Line	£12.08	29.8%	£9.54	24.6%
Additional Loss for Second Grid Line	£1.63	3.2%	£1.22	4.5%
Total Loss for Both Grid Lines	£13.72	33.1%	£10.76	29.1%
Loss for Telegraph Poles	£5.74	15.6%	£4.58	11.7%
Basic Wind Farm Average Loss	£5.56	8.7%	£6.90	18.0%
Extended Wind Farm Average Loss	£6.64	12.8%	£7.41	23.2%

Loss of Value by Age, Gender and Purpose

Table 6-28, Table 6-29 and Table 6-30 show the loss of value by age, gender and purpose.

Table 6-28 Loss of Values to US Tourists by Age

	Loss	Loss %	N
16 - 25	-£0.15	-1.5%	12
26 - 45	£5.47	4.9%	48
46 - 65	£7.02	15.5%	39
Over 65	£9.61	18.4%	4
Total	£5.56	8.7%	103

As with the UK example, the young appear to find the scenery equally attractive with or without turbines. In the US case, however, the loss for the elderly is greater than for any other group. Care, however, must be exercise because of low numbers in the category responding.

Table 6-29 Loss of Values to US Tourists by Accommodation

	Loss	Loss %	N
Hotel	£6.23	7.9%	70
Bed and Breakfast, Hostel	£4.24	10.9%	30
Other	£3.22	6.3%	3
Total	£5.56	8.7%	103

Table 6-30 Loss of Values to US Tourists by Activity

	Loss	Loss %	N
To see Scotland	£4.78	5.9%	68
Other	£9.41	16.5%	35
Total	£5.56	8.7%	103

There is no real difference in loss by accommodation type and, by implication, by income. There is no obvious explanation for the higher figure for Other activities except that it is paralleled to a lesser extent in the UK. The difference is not statistically significant (t=0.669 and 1.186)

Range of Values

As discussed under UK Results the variability within the sample is so large that it is difficult to find any statistically significant results. For the US sample this is illustrated in Table 6-31 which identifies the percentage of responses that indicated a loss, indifference (zero value change) and gain.

Table 6-31 Distribution of Values by site

	Negative	Neutral	Positive
Braes Current	57.30%	33.00%	9.70%
Braes Extended	68.00%	21.40%	10.60%
Additional Loss from Extension Braes	35.00%	44.60%	20.40%
Single Grid Line	80.60%	9.70%	9.70%
Double Grid Line	80.60%	9.70%	9.70%
Additional Loss from Extra Pylon	37.90%	46.60%	15.50%
Waterfall	59.20%	29.10%	11.70%
Spey	46.60%	35.90%	17.50%
Thurso Current	40.80%	42.70%	16.50%
Thurso Extended	48.50%	36.90%	14.60%
Additional Loss from Extension Thurso	30.10%	46.60%	23.30%

An important feature of this table is the level of indifference between the basic wind farm and the extension. Even in the case of the second pylon line, indifference exceeds negative reaction. This finding is in line with both the intercept study and the literature, a large group of people simply do not care.

6.10.3 US Tourist Perceptions

The perceptions of tourists form the US are similar to those from the UK but even more inclined to believe that there is a wind farm around each bend. There is some recognition that a tourist is less likely to see a wind farm in a scenic area but even here over 70% believe that they are likely, quite likely or very likely to see a wind farm.

Table 6-32 Views on likelihood of seeing a wind farm

	Not Scenic			Scenic
	N	Percent	UK Not Visited	N	Percent	UK Not Visited
Very likely	11	10.7%	8.9%	11	10.7%	6.5%
Quite likely	38	36.9%	16.6%	27	26.2%	26.0%
Likely	37	35.9%	38.5%	34	33.0%	39.1%
Not very likely	15	14.6%	33.7%	28	27.2%	23.7%
Not at all likely	2	1.9%	2.4%	3	2.9%	4.7%
Total	103	100%	100%	103	100%	100%

The effect of this belief is small. Fewer individuals say they would avoid areas with lots of wind farms and only 1 respondent identified it as a reason for not going to Scotland.

Table 6-33 Response of US visitors to Wind farms

	Frequency	Percent
Go to see them	37	35.9
No response	54	52.4
Avoid the areas	11	10.7
Avoid Scotland	1	1.0
Total	103	100

Far more would appear to want to go to an area to visit a wind farm.

6.11 Summary and conclusions

The internet study was designed and extensive pilots run using SNAP Software. It was then transferred to a commercial company GMI- MR for distribution to 600 randomly selected individuals from the UK and 100 from the US. The process was remarkably smooth and GMI- MR returned the data in SPSS format within the week. We would strongly recommend this type of surveying for similar projects.

The analysis showed that tourists, both domestic and foreign placed a value on a view from a bedroom in excess of £20 per room. This value was seriously eroded by wind turbines, pylons and telegraph poles. The pylons, in particular were disliked by virtually all with a mean loss of over £10 for UK tourists and over £13 for US tourists. Wind farms generated a loss between £7 and £8 for the UK and between £5 and £6 for the US.

The only distinctively different group were the young, who, in general were less worried than their parents.

The significance of age generated the hypothesis that families with children might have more appreciation of wind farms as a positive holiday experience. This was tested and the results shown in Table 6-34.

Table 6-34 Effect of children in the party

	UK (excl. Scots)		US
	Mean	N	Mean	N
No Children	£8.05	306	£6.72	72
Children	£6.32	132	£2.88	31

Whilst the US sample showed a difference, albeit not significantly different, this was not replicated in the UK sample.

As a general rule the further the tourist was away from Scotland the more they believed wind farms were more extensive than they actually are and the less they apparently minded. One marked feature was a failure to recognise that permission for developments in "highly scenic areas" are not normally allowed. There is an argument for either more National Parks or for a rigorous marketing of the concept of a National Scenic Area.

A substantial minority would either avoid an area or Scotland all together if the number of wind farms increases substantially. It is difficult to know what is meant by an area in this context and we prefer the findings of the intercept study because:

• Most respondents had just seen a wind farm
• The meaning of area was defined and explained to the respondents

The conclusions are that:

• The internet survey was effective and fast once linked to a commercial organisation.
• Scenery clearly has value.
• Wind turbines do reduce the value of the scenery although for a substantial proportion there is no loss of scenic value.
• The analysis suggests similar responses by nationality, age, gender, general expenditure, although there is some evidence that the young and children are indifferent.
• An estimate of the value lost is between a maximum of some 23.2% of the room price ( UK values only for extended farms) and a minimum .of 17.1% (wind-farm basic 90% UK, 10% US ) with a mean of 19.7%.. Taking into account the substantial individual variance into account our confidence range would be between 15% and 25% and these form the bounds for our sensitivity analysis.

6.12 Bibliography Chapters 5 and 6

Adobe 2002, Photoshop, 7.0th edn, Adobe, US.

Alberini, A. 2006, Handbook on Contingent Valuation, 1st edn, Edward Elgar, UK.

Alonso, F. 2002, "The benefits of building barrier-free: a contingent valuation of accessibility as an attribute of housing", European Journal of Housing Policy, vol. 2, no. 1, pp. 25.

Alvarez-Farizo, B. & Hanley, N. 2002, "Using conjoint analysis to quantify public preferences over the environmental impacts of wind farms. An example from Spain.", Energy Policy, vol. 30, no. 2, pp. 107-116.

Arrow, K., Solow, R., Portney, P.R., Leamer, E.E., Radner, R. & Schuman, H. 1993, Report of the NOAA Panel on Contingent Valuation, 1st edn, Federal Register, U.S.

Barreiro, J., Sánchez, M. & Viladrich-Grau, M. 2005, "How much are people willing to pay for silence? A contingent valuation study", Applied Economics, vol. 37, no. 11, pp. 1233-1246.

Bateman, I.J. & Willis, K. G. (Eds.) 1999, Valuing Environmental Preferences: The Theory and Practice of the Contingent Valuation Method in the US, EU and Developing Countries. 1st edn, Oxford University Press, Oxford.

Bateman, I.J. & Langford, I.H. 1997, "Non-users' Willingness to Pay for a National Park: An Application and Critique of the Contingent Valuation Method", Regional Studies, vol. 31, no. 6, pp. 571-582.

Bennett, R.M., Tranter, R.B. & Blaney, R. J. P. 2003, "The Value of Countryside Access: A Contingent Valuation Survey of Visitors to the Ridgeway National Trail in the United Kingdom", Journal of Environmental Planning and Management, vol. 46, no. 5, pp. 659-671.

Bjornstad, J. & Kahn, J. R. (Eds) 1996, The Contingent Valuation of Environmental Resources: Methodological Issues and Research Needs. Brookfield, Vt edn, Edward Elgar, Cheltenham, UK.

Boyle, K.J. & Bishop, R.C. 1984, Lower Wisconsin River Recreation: Economic Impacts and Scenic Values, Staff Paper Series edn, University of Wisconsin-Madison, Agricultural Economics.

Braden, J.B. & Kolstad, C. D. (Eds) 1991, Measuring the Demand for Environmental Quality. 1st edn, North-Holland/Elsevier, Amsterdam.

Brandolini, S. M. D. A. 2004, Economic Valuation of a Natural Area: the Southern Beach of Lido di Dante (Ravenna), University of Bologne.

Brookshire, D.S., Thayer, M.A., Schulze, W.D. & d'Arge, R.C. 1982, "Valuing Public Goods: A Comparison of Survey and Hedonic Approaches", The American Economic Review, vol. 72, no. 1, pp. 165-177.

Cameron, T.A. & Quiggin, J. 1994, "Estimation using contingent valuation data from a dichotomous choice with follow-up questionnaire.", Journal of Environmental Economics and Management, vol. 27, no. 3, pp. 218-234.

Carson, R.T. 1985, Three Essays on Contingent Valuation (Welfare Economics, Non-Market Goods, Water Quality)., University of California, Berkley.

Carson, R.T., Flores, N.E., Martin, K.M. & Wright, J.L. 1996, "Contingent Valuation and Revealed Preference Methodologies: Comparing the Estimates for Quasi-Public Goods.", Land Economics, vol. 72, no. 1, pp. 80-99.

Carson, R.T., Mitchell, R.C., Hanemann, M., Kopp, R.J., Presser, S. & Ruud, P.A. 2003, "Contingent Valuation and Lost Passive Use: Damages from the Exxon Valdez Oil Spill.", Environmental and Resource Economics, vol. 25, no. 3, pp. 257-286.

Cummings, R.G., Brookshire, D.S. & Schulze, W. D. (Eds) 1986, Valuing Public Goods: The Contingent Valuation Method. 1st edn, Rowman & Allanheld, Totowa, NJ.

Davis, R. 1963, "Recreation planning as an economic problem.", natural resources Journal, vol. 3, no. 2, pp. 239-249.

Diamond, P.A. & Hausman, J.A. 1994, "Contingent Valuation: Is Some Number better than No Number", The Journal of Economic Perspectives, vol. 8, no. 4, pp. 45-64.

Fix, P.J. & Manfredo, M.J. 2005, "Divergent Validity of Contingent Valuation Responses in a Wildlife-Related Application", Human Dimensions of Wildlife, vol. 10, no. 4, pp. 239-248.

Green, D., Jacowitz, K.E., Kahneman, D. & McFadden, D. 1998, "Referendum contingent valuation, anchoring, and willingness to pay for public goods.", Resource and Energy Economics, vol. 20, no. 2, pp. 85-116.

Green, C.H. & Tunstall, S.M. 1991, "The evaluation of river water quality improvements by the contingent valuation method", Applied Economics, vol. 23, no. 7, pp. 1135.

Haab, T.C. & McConnell, K., E. 2003, Valuing Environmental and Natural Resources: The Econometrics of Non-Market Valuation, Paperback edn, Edward Elgar Publishing, Inc., UK.

Hanemann, M., Loomis, J. & Kanninen, B. 1991, "Statistical Efficiency of Double-Bounded Dichotomous Choice Contingent Valuation.", American Journal of Agricultural Economics, vol. 73, no. 4, pp. 1255-1263.

Hanley, N., Shaw, W.D. & Wright, R.E. (eds) 2003, The New Economics of Outdoor Recreation, 1st edn, Edward Elgar, Cheltenham, UK.

Hanneman, W.M. 1985, "Some Issues in Continuous and Discrete-Response Contingent Valuation Studies.", Northeastern Hournal of Agricultural Economics, vol. 14, no. 1, pp. 5-3.

Heywood, I., Cornelius, S. & Carver, S. 2002, An Introduction to Geographical Information Systems, 2nd edn, Pearson Prentice Hall, Essex, UK.

Kanninen, B.J. 1995, "Bias in discrete response contingent valuation.", Journal of Environmental Economics, vol. 28, no. 1, pp. 114-125.

Loomis, J.B. 1990, "Comparative reliability of the dichotomous choice and open-ended contingent valuation techniques", Journal of Environmental Economics and Management, vol. 18, no. 1, pp. 78-85.

McFadden, D. & Leonard, G. (eds) 1993, Issues in contingent valuation of environmental goods: methodologies for data collection and analysis, in Contingent Valuation a Critical Assessment (Ed.) Hausman, J. A., 1st edn, North-Holland Publishing Co, Amsterdam.

Mitchell, R.C. & Carson, R.T. 1989, Using Surveys to Value Public Goods: The Contingent Valuation Method. 1st edn, Resources for the Future, Washington D.C.

SNAP 2007, SNAP Surveys, Professional Edition 9.0 edn, SNAP, London.

SNH 2006, Visual Representation of Wind farms: Good Practice Guide, Horner + Maclennan and Envision, Inverness.

Sparkes, A. & Kidner, D. 1996, 1996-last update , A GIS for the Environmental Impact Assessment of Wind Farms [Homepage of ESRI], [Online].
Available: http://gis.esri.com/library/userconf/europroc96/PAPERS/PN26/PN26F.HTM [2007, 05/05] .

Treiman, T. & Gartner, J. 2006, "Are residents willing to pay for their community forests? Results of a contingent valuation survey in Missouri, USA", Urban Studies, vol. 43, no. 9, pp. 1537.

Yoo, S., Shin, C. & Kwak, S. 2006, "Inconvenience cost of spam mail: a contingent valuation study", Applied Economics Letters, vol. 13, no. 14, pp. 933.