Value of bathing waters and influence of bathing water quality: final research report

3. Methodology

3.1 Introduction

Given that the aim of the research was to investigate both economic and wellbeing benefits of bathing waters, this research project used a mixed methods approach to collect quantitative and qualitative primary data. The fieldwork component of the research was conducted at five bathing water sites across Scotland in the summer and early autumn of 2017.

3.2 Research questions addressed and overall research approach

The following research questions ( RQ) were used to focus the research:

• RQ1 What are the range of benefits (economic, health, social and cultural) of bathing waters at the local and national levels, and how can these be measured?
• RQ2: What value do people put on information about bathing water quality (signs and signage), how do they understand that information (particularly information about changes in classification) and how does it influence beach visit decisions?
• RQ3: How can the benefits (or costs) of an improvement (or deterioration) in bathing water quality classification be assessed, taking account of effects on the national as well as local economies and both short- and long-term effects?
• RQ4: What learning can be derived from the evidence about the management and assessment of designated bathing water sites, and the overall value of bathing water quality in Scotland?

To fully address these research questions a mixed-methods approach was developed for the project and, as such, each chapter on the findings reports from across the range of data sources used (quantitative and qualitative). The project addresses the issue of the value of bathing water from a number of perspectives. Figure 3.1 shows the relationship between those value perspectives and the type of data collected and analysis that was carried out.

Two fieldwork methods were used across five bathing water sites of varying bathing water ( rBWD) classifications together with an online survey to a nationally representative sample of the Scottish population. The intention was to provide a mix of quantitative and qualitative evidence about perceptions of bathing water quality, including:

• Response to different forms of information provision;
• The range of economic, cultural, and health and wellbeing benefits provided by beach visits; and
• Potential changes in attitudes and behaviour in response to changes in bathing water quality.

Figure 3.1: Relationship between different perspectives on the value of bathing water and methods for their investigation

The fieldwork methods used were as follows:

1. An onsite survey of over 500 visitors (in total) to the five bathing water case study sites across Scotland;

2. Six focus groups in the five bathing water case study sites with community members, businesses and one specific user group (wild-swimmers); and

3. An online survey of 1,000 representative of the Scottish population to gather their preferences for bathing water quality at the site they visit most often (users) or most associate with (non-users) and for the characteristics of all bathing water sites in Scotland.

The methods provided complementary information to ensure that the research objectives could be met. The five chosen case study sites are listed below and shown geographically along with their current (2015/16) rBWD status on Figure 3.2.

1. Nairn (Central), Highlands;

2. Ayr (South Beach), South Ayrshire;

3. Troon (South Beach), South Ayrshire;

4. Portobello (West), City of Edinburgh; and

5. Gullane, East Lothian.

Figure 3.2: Location and current (2015/16) rBWD status of the five case study bathing waters considered in the research (onsite survey and focus groups)

The locations were chosen to represent a range of bathing water quality ( rBWD) classifications (10-year average) and site types, as shown in Table 3.1 below. The bathing water site types were intended to be a pragmatic categorisation, identified through discussion with the Steering Group as follows: (i) ‘coastal resorts’ are locations with resort features / amenities (e.g. promenades, large caravan sites); (ii) ‘coastal towns’ are locations where the bathing water site is closely connected to a larger coastal settlement; and (iii) ‘coastal villages’ are sites linked to smaller coastal settlements that can be readily classified as villages ^[10] .

Table 3.1: Rationale for study site selection

rBWD status [11]	Coastal resort	Coastal town	Coastal village
Poor	Ayr (South Beach)	Nairn (Central) Portobello (West)
Sufficient
Good	Troon (South Beach)
Excellent			Gullane

3.3 Data collection methods

3.3.1 Onsite survey

The aim of the onsite survey was to understand more fully the purpose and benefits of bathing water visits, including information about the visitors (age, gender, etc), their expenditure and activities undertaken, and how these would change if bathing water sites failed to meet sufficient status. Responses more particularly relevant to the econometric analysis included information about respondent’s choice of beach (including substitutes considered) and their journey to the beach, including the mode of travel, the distance they travelled, the time it took, where they started their journey, and how often they visit. These responses were then used within a trip generating function ( TGF) and individual travel cost model ( ITCM) which examine how the number of visits an individual makes to a site changes as travel costs (distance and time) change, whilst other factors that would influence the choice are kept constant (such as substitute sites, household size, age, and household income). This is interpreted as the WTP for access to the sites, and a minimum indication of how valuable the benefits of a visit are perceived to be (based on the common economic assumption that if the benefits were perceived to be less than the costs, the activity would not be undertaken).

The onsite survey questionnaire was developed in collaboration between eftec, CEP, and the Scottish Government. It was based on the survey design of previous large-scale bathing water research undertaken by eftec. In order to maximise the level of participation in the survey at each case study site , the questionnaire length was kept to around 15 minutes. The content of the questionnaire is set out in Table 3.2 below. The full onsite survey instrument is included at Annex 6 in the standalone technical annexes volume.

Showcards were used to present respondents with detailed information, where necessary, and the range of possible response options, where relevant. Annex 2 in the standalone technical annexes volume sets out details of the application of the onsite survey data in the estimation of visitor expenditure and gross value added ( GVA), the trip generating function ( TGF) analysis and travel cost models ( TCM).

Interviews were conducted in-person at the five case study sites . In total 516 interviews were conducted, spread evenly across the locations (between 102 – 105 at each site).

In advance of the interviews, the project team liaised with local contacts and beach managers to ensure relevant local stakeholders were informed and permission was sought where appropriate.

Table 3.2: Onsite survey content

Questionnaire section	Data collected
Visitor profile	Type of visit, party composition, purpose of visit, accommodation, travel and expenditure
Details of beach usage	Time spent at beach, activities undertaken
Attitudes towards water quality and cleanliness	Perception of quality of beach, perceived restorativeness, awareness of information signs and designation
Visits to other Scottish beaches	Frequency, activities undertaken, distances travelled
Future visit behaviour	Likely response to BWQ advisory signs
Demographics and geographic information	Postcode, household income, age, gender, ethnicity

Survey respondents were selected using a random probability approach in which every ‘nth’ person at the location was approached for interview. This approach ensures the sample of survey respondents is broadly representative of the visitors to the beach during the period of time that the interviewer is present. The interviews were conducted between July and September 2017; during weekdays and weekends and both term and school holiday times. This covered a useful spread of times and conditions over which visitation may differ, as interviews were conducted regardless of the weather. The study did not intend to provide a comprehensive year-round perspective but focused on the bathing water season. Interviewing was conducted using a computer aided personal interview ( CAPI) methodology. Each interviewing shift lasted six hours.

Response data was processed and analysed. The results are reported on their own and are inputted to the local economic impact analysis.

3.3.2 Online survey

The purpose of the online survey was to provide a nationally representative sample to understand the purpose and benefits (for both users and non-users) of bathing water visits, and how these would change if bathing water sites failed to meet sufficient status.

The online survey also more explicitly explored the preferences of individuals for different environmental quality outcomes through a stated preference approach using a discrete choice experiment ( DCE). An example is shown in Figure 3.2 below. The scope of the online survey was wider: the onsite survey was only about the visit to the beach during which the interview took place, while the online survey covered the beach the respondent visits most often (users) or most associated themselves with (non-users) as well as all bathing waters in Scotland more generally.

Figure 3.3: Survey choice card used in online survey

In particular, respondents were required to select their preferred option from three alternatives that traded-off changes in:

1. The total number of bathing waters in Scotland failing to meet ‘sufficient’ status (% bathing waters failing);

2. The bathing water status of the beach they visit most often (users) or that they feel the most association with (non-users) (‘poor’, ‘sufficient’, ‘good’ or ‘excellent’);

The cleanliness of the beach they visit most often or feel most association with (% litter removed); and

3. The cost to their household to secure these improvements (paid through increased water bill – which is a ‘payment vehicle’ used in previous bathing water related surveys).

Option C in Figure 3.3 is the ‘current situation’ that does not involve respondents paying anything more than they currently do.

A (stated) choice experiment uses this process to derive estimates of respondents’ preferences for the various improvements, by presenting respondents with repeated choices across experimentally designed alternative bundles of improvements, and asking them to choose their most preferred bundle from the available set in each repeated choice. When one of these characteristics is the cost of the bundle and customers choose one bundle (a specified package of improvements) over others, they implicitly reveal their trade-off between their income (money) and the single improvements included in each bundle in their choice set. Such a trade-off is the marginal "willingness to pay" ( WTP) value of that characteristic of the bundled good (i.e. the value of a one unit change in provision of an improvement

The choices of respondents can then be used to estimate marginal WTP for changes in these outcomes. This complements the onsite survey by providing a nationally representative sample, information from both users and non-users of bathing waters and marginal values for differing levels of environmental quality.

Table 3.3: Online survey content

Questionnaire section	Data collected
Visitor profile	Type of visit, party composition, purpose of visit, accommodation, travel and expenditure
Details of beach usage	Time spent at beach, activities undertaken
Attitudes towards water quality and cleanliness	Perception of quality of beach, perceived restorativeness, awareness of information signs and designation
Visits to other Scottish beaches	Frequency, activities undertaken, distances travelled
Attitudes toward levels of litter, site-specific and nation-wide sites with failing BWQ	Perceived cleanliness of beach most visited / most associated with, preferences for levels of reduction in litter at the site, and preferences for differing levels of beaches in Scotland with ‘poor’ status
Future visit behaviour	Likely response to BWQ advisory signs
Demographics and geographic information	Postcode, household income, age, gender, ethnicity

The development of the online survey questionnaire followed that of the onsite survey, with input from the project team and the Scottish Government, and consideration given to keeping a reasonable questionnaire length (~20 minutes). There were some differences, however, in the content of the online survey which can be seen in Table 3.3. The full online survey instrument is included at Annex 6 in the standalone technical annexes volume.

The online survey was run during October and November 2017 using an online panel of the populations of Scotland maintained by Survey Sampling International ( SSI), a market research company. Sampling quotas for gender, age, and socio-economic group were used to ensure a nationally representative sample of 1,013 respondents.

Data from the responses were processed and analysed. Annex 3 in the standalone technical annexes volume sets out the details of this analysis.

3.3.3 Focus groups

The focus groups allowed further exploration of people’s attitudes and values towards bathing water quality classifications and water quality information on beaches, focussing on residents and business people in the five case sites (although noting that six focus groups were held, as explained below).

A mixture of business and community views were collected, to allow a comparison between these perspectives, with local business focus groups held at Nairn and Ayr, and local community focus groups at Troon, Gullane, and Portobello. Two focus groups were held in Portobello: (i) a local community session as was the case in Troon and Gullane; and (ii) a specific session with a group of wild swimmers who swim off Portobello. Community participants were recruited through community councils and other local organisations, social media channels, advertisements and direct approaches. The online platform Survey Monkey was used to facilitate sign up. Business participants were sought through the local authority, business associations and direct approaches. A small monetary incentive (£30) was provided for all participants.

The focus groups involved 4-11 participants (6 on average). The profile of the participants for all the focus groups can be found in Annex 8 in the standalone technical annexes volume. The community focus groups participants were mainly women (85%), with a range of ages, and employment statuses. The business focus group participants were more balanced in terms of gender (61% female, 39% male) comprising a mix of businesses (in terms of types and size).

Three additional telephone interviews were conducted after the focus group meetings with businesses that had been unable to attend (two in Ayr and one in Nairn) to provide a wider range of input.

The focus group schedule (see Annex 7) was developed to explore issues emerging from the results of the two surveys, and included a scenario exercise which compared participants’ expectations of future visits to the beach, based on two different scenarios:

1. In the first future scenario the water quality of the local site had improved and participants were asked if that changed what they might do and feel, and if it might change opportunities for their communities; and

2. In the second future scenario the water quality of the local site had deteriorated and again participants had to comment on how that might affect their actions and feelings, together with an assessment of impacts on their local communities.

The sessions were facilitated by two members of the project team, one of whom took notes throughout, to capture detailed information about the process, while the facilitator introduced topics and prompted discussion. A pilot was conducted in Troon and was assessed as having achieved the desired outputs. Therefore, results were merged with data from the subsequent focus groups.

3.4 Analytic approach

The analysis adopted a mixture of quantitative and qualitative approaches, which are outlined in more detail in the following sub-sections.

3.4.1 Quantitative analysis methods

The sub-sections below introduce the quantitative analytical tools used in the research.

Descriptive statistics

Descriptive statistics (%) for all the questions on both surveys (onsite and online) were carried out. Details of the demographics (e.g. age, gender, socio-economic group ( SEG), ethnicity, employment status) for all participants can be found in Annexes 2 and 3. Descriptive statistics for all the questions can be found in Annex 5. The specific analytic statistics used for different aspects of the survey data are discussed in the following paragraphs.

Revealed preference method

Revealed preference methods use data about individuals’ behaviour in an actual market to estimate the value they place on goods and services that are not marketed. In this project, a travel cost approach was adopted using data on expenditure an individual or household group makes for and during a visit to a bathing water site as a minimum proxy of the benefit they receive from the visit. The underlying assumption is that people would not spend more to visit a site than the enjoyment they get from the site, even if the latter is not in monetary terms.

The onsite survey elicits information from respondents about their visits to bathing waters, including the: frequency of visits; activities undertaken; and visitor spend. The use of visitor spend data feeds into the local economic impact assessment (discussed further in the following sub-section). The survey also captures information regarding how (and if) visitors would change their behaviors if they saw a sign advising against bathing due to poor water quality. A so-called trip-generating function ( TGF) is created using this information to calculate the number of visits to a bathing water site by an individual per year as determined by a series of explanatory factors, including the availability of substitute sites. These factors are summarised and discussed in Annex 2 in the standalone technical annexes volume.

If the TGF can successfully predict the number of visits currently, it can also be used to predict future visits under different circumstances of the important factors. The function is applied to estimate the reduction in visitor numbers based on the respondents’ reaction to advisory signs being displayed. The loss of visitor spend in the local area and recreational value due to a reduction in visits can then be estimated.

The TGF is based on the specification of a model presented in Annex 2 Table 10. The overall model fit for the results shown in Annex 2 Table 10 is good (R ² = 0.617) and the findings are valid given that they are in line with prior expectations.

The monetary value of recreation visits (£/visit) – in terms of WTP for access and recreation opportunities – is estimated via an individual travel cost model ( ITCM). As previously mentioned ( section 3.3.1), the WTP is interpreted as a minimum indication of how valuable a visit is to the visitors (based on the common economic assumption that if the benefits were perceived to be less than the costs, the visit would not be undertaken). This is calculated as the current (baseline) or reduction in the number of trips to a site multiplied by the value per trip measured in terms of travel costs. The ITCM examines how the number of visits an individual makes to a site changes as travel costs (distance and time) change, whilst controlling for substitute sites, overnight visitors, total expenditure, age, and household income. The travel cost elements apply the following assumptions for distance and distance to substitute sites: travel time is valued at 75% of the average wage rate and fuel cost at £0.30/mile.

Analysis of local economic impacts

Local economic impact analysis was conducted to add to the quantitative evidence exploring the range of economic benefits provided by beach visits. Expenditure data was gathered through the onsite survey for the following broad spend categories associated with bathing water / beach usage:

• Eating and drinking in cafes, pubs, restaurants;
• Buying food, drinks or snacks from shops;
• Shopping such as souvenirs and items for the beach;
• Tourist activities such as local attractions or water sports lessons;
• Travel and transport, for example fuel or train tickets;
• Car parking; and
• Accommodation.

The data collected from the onsite survey on expenditure, group size and mode or transport was supplemented by local and site-specific data. The most relevant supporting information for the economic impact modelling relates to estimates and proxies of total visitors to bathing waters including visitor data provided by SEPA, data on car parking capacity, and data on accommodation capacity.

Overall the analysis features four discrete steps:

1. Estimation of the number of annual visitors to sites using available tourism data;

2. Estimation of expenditure per site visitor per day based on the onsite survey results;

3. Estimation of total visitor spend per year for each site from steps 1 and 2; and

4. Application of the Cambridge Model to generate:
a. Estimates of business turnover supported;
b. Estimates of associated employment supported by expenditure; and
c. Estimates of gross value added.

Further information on the local economic impact analysis, including key assumptions, is provided in Annexes 3 and 4 in the standalone technical annexes volume.

Stated preference method

Based on analysis of the choice experiment responses, values for improvements in bathing water quality and beach characteristics, in terms of WTP for these improvements, were estimated. Analysis of choices made by respondents reveals the premium they are willing to pay for incremental improvements in each attribute. Technically WTP for an attribute is estimated by dividing the coefficient for that attribute by the cofficient of the cost attribute.

Further details of the estimation strategy, including the model specification that provided the best fit to the data and model results, are provided in Annex 3.

Analysis of restorative benefits

In order to analyse the factors associated with perceptions of the restorative benefits of bathing waters and the specific role of BWQ within this, regression analyses were carried out on both the onsite and online data. Factors included in the analyses were drawn from previous research (e.g. Wyles et al., 2016). The variables included in the analysis are listed below:

• Age;
• Gender;
• Perceived BWQ;
• Mode of transport;
• Companions;
• Distance from home;
• Frequency of visits to this beach; and
• Purpose of visit.

To examine the effect of a potential deterioration in BWQ on perceptions of restorativeness, t-test and ANOVAs were carried out. The t-test tested whether if the water quality deteriorated it would affect the restorative potential of the site by comparing respondents’ views on the current status with those if it deteriorated to ‘poor’ (advisory against bathing). An ANOVA was carried out to see what influences the relationship between BWQ and perceived restorativeness, comparing those who did a water activity with those who did not.

3.4.2 Qualitative analysis method

An audio recording was made of all focus group sessions, to allow transcription and thematic analysis. In analysing this data, an inductive (bottom-up) as well as a deductive (top-down) thematic approach was used. This involved coding the data according to themes that appeared in the focus group transcripts (inductive) or were included in the focus group schedule (deductive). This allowed answers to the questions posed, but also prevented constraining the data by only looking at already named categories.

The software package Dedoose ^[12] allowed codes to be added to transcript text, and then for those excerpts to be exported separately, to enable a further exploration of specific themes. Focus group findings were summarised, their main themes deduced, and linkages within and across focus group discussions explored in relation to the research questions. Using Dedoose allowed an identification of codes that appear most frequently, and this was used as a pragmatic means of prioritising analysis.