Annex 2: Econometric analysis of onsite survey data – revealed preference method
This annex provides selected technical detail and results from the econometric analysis of the survey data to support the findings presented in the main report. This annex is structured as follows:
- Onsite survey sample profile ( section A-2.2); and
- Estimation of trip generating function ( section A-2.3).
A-2.2 Onsite survey sample profile
The onsite survey was designed to include a representative sample of beach visitors, rather than a nationally representative sample. This is particularly helpful in terms of identifying who visits Scotland’s beaches in terms of gender, age, income and other socio-demographic factors, and estimating how actual visitors value beaches. However, this also means that the results cannot necessarily be assumed to be compatible with national data, or representative of the national population.
This sub-section presents an overview of the onsite survey sample. Figures are rounded to nearest whole number which may cause totals to sum to more than 100%. The onsite survey consists of 516 respondents. Sample sizes are indicated within each of the subsequent tables. Where the indicated sample size differs from the overall survey sample size, this is explained by the routing of the surveys (see accompanying survey documents in Annex 6). Full summary stats for each question of the surveys are provided as accompanying excel documents to the report.
Table A2-1: Onsite survey location
|Ayr (South Beach)||105||20%|
|Troon (South Beach)||104||20%|
|Other (Please specify)||0||0%|
Location (drawing on data from onsite survey question 1)
Table A2-2: Onsite survey visitor type
|Travelled to the area for work||0||0%|
|On a short break, away from home for 2-6 days||36||7%|
|On a long break, away from home for 7 or more days||35||7%|
|Foreign visitor on holiday||20||4%|
Visitor type (drawing on data from onsite survey question 2)
Table A2-3: Onsite survey number of people in group
|Number of people in group||Adults
|More than 10||2||0%||4||1%|
Number of people in group (drawing on data from onsite survey question 9)
Table A2-4: Onsite survey gender
Gender (drawing on data from onsite survey question 37)
Table A2-5: Onsite survey age range of respondent
|Age range of respondent||N||%|
|Prefer not to say||2||0%|
Age range of respondent (drawing on data from onsite survey question 38)
Table A2-6: Onsite survey annual total household income
|Annual total household income||N||%|
|Less than £10,000||17||3%|
|£91,000 - £100,000||1||0%|
|Prefer not to say||265||51%|
Annual total household income (drawing on data from onsite survey question 39)
Table A2-7: Onsite survey ethnic group
|Gypsy or Irish Traveller||0||0%|
|Any other White background||11||2%|
|White and Black Caribbean||0||0%|
|White and Black African||3||1%|
|White and Asian||2||0%|
|Any other mixed background||0||0%|
|Any other Asian/Asian British background||0||0%|
|Any other Black/African/Caribbean background||0||0%|
|Any other background||8||2%|
|Don't wish to say||0||0%|
Ethnic group (drawing on data from onsite survey question 41)
A-2.3 Estimation of trip generating function
A-2.3.1 Factors influencing the frequency of visits
Figure A2-1 summarises the average distance travelled in miles by visitor type. Stated distance and travel time are found to be highly correlated and consistent with visitor type with day trippers travelling notably further than visitors who are local residents. In considering respondents’ mode of transport, Figure A2-2 presents results which show respondents that walked, jogged or travelled by bike travelled shorter distances on average than respondents with another mode of transport.
Figure A2-1: Average distance travelled in miles by visitor type (drawing on data from onsite survey questions 26 and 27)
Figure A2-2: Average distance travelled by respondents’ mode of transport (drawing on data from onsite survey questions 26 and 27)
Respondents were asked how many trips they made to a beach in a year. Table A2-8 presents a breakdown of the average distance travelled across the range of trips made by respondents. Respondents who make fewer trips to a beach in a given year tend to travel further distances and vice versa.
Table A2-8: Average distance travelled by respondents’ visit frequency
|Number of visits per year||Average distance (miles)||Sample size|
|More than 100||3||84|
|Less than 3 trips||25||211|
Average distance travelled (miles) and sample size (n) of respondents by number of visits per year (drawing on data from onsite survey questions 27 and 4).
A-2.3.2 Trip generating function
A trip-generating function ( TGF) was specified to analyse the factors influencing visits to bathing waters. The TGF predicts the number of visits by an individual to a beach in a given year. It combines data from the onsite survey for the observed number of visits per year and the stated reduction in visits due to an advisory against bathing. This allows the function to be applied to estimate the reduction in visitor numbers to a site as a result of advisory signs being displayed. In particular, the TGF is based on the stated behaviour (‘contingent behaviour’) questions which elicit the change in visitor behaviour in the event of signs advising against bathing being displayed. This allows for the change in visits to be estimated.
This analysis also considered the categorization of sites by type, whether the site is a rural beach, a coastal village/town beach or a resort beach. The category of each is presented in Table A2-9 below.
Table A2-9: Site type
|Ayr (South Beach)||Coastal resort|
|Nairn (Central)||Coastal town|
|Portobello (West)||Coastal town|
|Troon (South Beach)||Coastal town|
Table A2-9: Site type of sample locations
The TGF is based on the specification of a ‘best fit’ model presented in Table A2-10. The full model predicts the number of trips per visitor per year to a bathing water as a function of a variety of explanatory variables including distance travelled, site characteristics, visitor characteristics and number of substitute sites ( Table A2-10).
Overall the main findings accord with reasonable prior expectations and the analysis shows, that, all else equal:
- The further the distance travelled to the site, the fewer the number of visits made;
- As distance to substitute sites increases, the number of visits to the site increases;
- Income is not a significant determinate of number of trips;
- A site with bathing water status of ‘sufficient’ are visited more often than ‘poor’;
- A site meeting ‘excellent’ status is not significant in impacting trip numbers;
- Day trippers tend to visit a given beach less often than local residents;
- Overnighters tend to visit a given beach less often than local residents;
- Male visitors visit less frequently than female visitors;
- Older respondents visit the beach more frequently;
- Dog walkers tend to visit a given beach more often than those who visit for other activities; and
- Coastal towns are visited more frequently than coastal resorts.
The overall model fit for the results is good (R2 (adj) = 0.617) and certain findings above are valid given that they are in line with prior expectation. For example, the results are as expected in terms of distance, that dog walkers visit more often than others.
These results also present a variety of interesting findings for which there were no prior expectations. From a social point of view that older respondents visit more frequently is worth highlighting. Some older aged persons are less likely to have alternative opportunities for recreation, and so implications of declining bathing water quality might disproportionately affect this population. Likewise, that income is not a significant factor in determining the number of visits highlights the value of bathing waters across the entire population, with potential implications of improving environmental equality.
In terms of bathing water quality, the results indicate that a site meeting sufficient status (i.e. not failing) impacts the number of trips people make, however, improvements to good or excellent status do not. This implies that individuals care about a site in terms of not being ‘poor’ however, are less concerned with improvements beyond that. Discussions with focus groups also support this finding.
There are of course limitations within this study and the methods used that should be considered. Although the number of bathing waters considered (five) is a small proportion of the total number of beaches across Scotland (86), there is added confidence in the results due to the survey being based on one that was used by the Environment Agency to assess the perceptions, behaviours, and values of visits and bathing water quality across 43 sites in England (sample size of approx. 10,000) (eftec et al., 2014). This previous work provides a useful reference against which results from this study can be compared/validated. The results from both studies in terms of the factors that are found to impact the number of visits, as well as the implications of bathing water quality on visits align. Because of this, there can be a higher degree of confidence in that we have included the most important factors in determining visits to bathing waters, and in terms of the relative lack of importance of bathing water quality.
However, adequate care should be taken when using the results of this study to explore values associated with other beached. For safe measure, the results should be interpreted as applying to similar bathing waters and similar types of visitors. For example, there may be certain types of beaches that offer a vastly different array of recreational opportunities, and would not necessarily be comparable to the sites selected within this study, It is therefore recommended that when applying the values produced from this study, Defra’s value transfer guidelines be used (eftec, 2009). These guidelines emphasise transparency and the appropriate use of sensitivity analysis to address concerns of accuracy.
Table A2-10: Trip-generating function – factors influencing the number of visits to a bathing water site per year (onsite survey)
|Explanatory variable||‘Best Fit’ model|
|Distance to beach (m)||-0.05268***|
|Distance to substitute (km)||0.007484***|
|Water quality ‘sufficient’ (from ‘poor’)||0.094004***|
|Water quality ‘excellent’ (from ‘poor’)||0.06653*|
|Day visitor (local residents as base)||-0.769842***|
|Overnight visitors (local residents as base)||-3.030816***|
|Site type (Coastal resort as base)||0.287084***|
|R – squared (adjusted)||0.617|
|Base (sample size)||462|
*** indicates coefficient estimate is statistically significant at the 1% level, ** indicates coefficient estimate is statistically significant at the 5% level, * indicates coefficient estimate is statistically significant at the 10% level. Distance travelled is coded in terms of miles (as respondents answered in miles in the survey); distance to substitute beaches is measured in kilometers (km) based on geographical information system ( GIS) calculations. The sample size has been reduced due to eliminating responses with missing (refused) postcode data.
A-2.3.3 Estimation of travel cost model
The individual travel cost model ( 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, household size, age, and household income. As discussed previously, this is interpreted as the WTP for access to the sites. Given the relatively small number of observations for each individual site, data are pooled to estimate WTP per visit, which gives a sample of around 500. The ITCM is estimated using a Poisson model. The travel cost elements apply conventional 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/m. Table A2-11 presents the regional ITCM parameter estimates.
Table A2-11: Individual travel cost model - factors influencing the number of visits to a bathing water site per year
|Travel cost per person per visit||-0.112407***|
|Travel cost per person per visit to substitute sites||0.002239|
|Day visitor (local residents as base)||-0.950565***|
|Overnight visitors (local residents as base)||-3.058914***|
|Water quality ‘sufficient’ (from ‘poor’)||0.089608|
|Water quality ‘excellent’ (from ‘poor’)||0.059104|
|Site type (Coastal town)||0.323574*|
|R – squared (adjusted)||0.601|
|Base (sample size)||462|
*** indicates coefficient estimate is statistically significant at the 1% level, ** indicates coefficient estimate is statistically significant at the 5% level, * indicates coefficient estimate is statistically significant at the 10% level. The sample size has been reduced due to eliminating responses with missing (refused) postcode data.
The WTP for access – shown in Table A2-12 below – is determined by the ratio of observed annual visits and the coefficient of travel costs.
Table A2-12: Estimated WTP for recreation visits (£/visit)
|‘Best fit model’||WTP (£/visit)|
There is a problem
Thanks for your feedback