2. Research methodology
Summary of research methodology
The research has been designed around a case study approach to understand the impacts of short-term lets (STLs) across different types of communities in Scotland. Five case study areas were selected to reflect different types of areas and experiences. The research was mainly exploratory, undertaken through qualitative methods with residents, hosts, community actors and local businesses. This was combined with short quantitative surveys of residents and hosts, primarily used to recruit participants for qualitative in-depth interviews, but also to provide insights into the possible scale of impacts. Secondary data analysis provided contextual information on the incidence of STLs across Scotland.
The research fieldwork was undertaken between June and September 2019. The methodology, and number of responses achieved are set out below.
Case study area selection
Drawing on secondary data analysis, five different areas were identified to represent types of communities and experiences of STLs. Selection was based on the diversity of communities, incidence and growth of STLs, rural and urban context, and the socio-economic profile of the area using the Scottish Index of Multiple Deprivation (SIMD). The study areas are not necessarily typical of Scotland as a whole, but they are likely to be representative of those areas where STLs are concentrated. The areas are as follows:
- Edinburgh (Central ward comprising the Old Town, New Town and Tollcross) – established world heritage site, city-centre tourism;
- The East Neuk of Fife (East Neuk and Landward ward coastal towns excluding St Andrews) – established coastal second home location;
- Fort William – rural expanding tourism;
- Glasgow (City Centre ward including Merchant City, Anderston and Yorkhill) – inner-city centre event tourism;
- Eilean a' Chèo (Skye) – remote rural established tourism.
Chapter 6 includes the profile of these areas and summary of key findings for each.
Scotland-wide analysis of Airbnb listings - secondary data analysis
The secondary data analysis undertaken for this research relates to Airbnb listings only. Airbnb does not constitute the entire STLs market, since some landlords do not use this platform to advertise their properties. However, the combined qualitative and quantitative research undertaken for this study suggests that Airbnb comprises a very substantial part of the STLs market.
Secondary data analysis of STLs in this report is based on Scotland's Airbnb listings from 19 May 2019. Data used in this project was sourced from Inside Airbnb, and then cleaned and filtered i.e. to ensure as far as possible only active listings were considered. The raw data file was filtered so that only 'active listings' were included in the analysis. For the purposes of this research, 'active listings' were considered to be those where the host had updated their availability calendar in the past six months. Full details of the methodology can be found in Appendix 1.
The analysis resulted in a total figure of 31,884 active listings for the whole of Scotland. By way of comparison, this figure is close to the total of 31,000 active listings for Scotland reported by Airbnb in their UK Insights Report from 2018. The distribution of the cleaned data was then examined, revealing significant geographic clustering. The presence of STLs was also compared to the total dwelling stock in Scotland, in order to derive an estimated penetration rate for individual locations. An analysis of hosts was also undertaken including the category of hosts and the number of listings by host.
Chapter 3 sets out the context of the incidence of STLs across Scotland.
Resident household survey and qualitative research
Small household surveys were undertaken with residents in the 5 case study areas. The survey served two purposes; to collect indicative quantitative information relating to the perceived positive and negative impacts of STLs and to identify potential interviewees for the residents strand of the qualitative research. The resident survey questionnaire is published separately. It is important to note that this feedback is not a representative sample of the population as a whole, but rather is reflective of views within the specific wards in the five case study areas. The survey was primarily conducted on a face-to-face basis, with a small amount of additional telephone-based recruitment being undertaken where required to augment the sample (12 out of the 197 surveys). A target was set to recruit 40 participants from each area, with quotas on specific case study location (by specific post code), gender and limits on the number of surveys in any one street or stair. The number of resident surveys achieved were 197 completed surveys broken down as 40 each in Glasgow and Skye, and 39 each in Edinburgh, Fife and Fort William. The profile of respondents is included in Appendix 2.
Residents recruited from the surveys were contacted for semi-structured telephone interviews. The discussion guide for these interviews is published separately. There was a target of 100 resident interviews with 20 in each of the five areas. A total of 100 resident interviews were achieved with a breakdown of 24 in Edinburgh, and 19 each in of the East Neuk of Fife, Fort William, Glasgow and Skye.
Host online survey and qualitative research
As with the residents, the host survey also served two purposes and was conducted online, with the survey questionnaire included in Appendix 3. Direct invitations to participate in the survey were issued in each of the case study areas by Airbnb, with two invitations issued. This was the main source of responses, representing 183 (81%) of the 227 responses. The balance of responses came from promotion of the survey on social media platforms in areas where initial response levels were lower (Airbnb host clubs, Skye rooms, Skye free ads) and responses to direct invitations issued by the consultancy team, where contact details for potential hosts could be accessed from publicly available sources. The profile of respondents, and summary of responses is included in Appendix 2.
Like the resident survey, hosts were recruited for semi-structured telephone interviews. The discussion guide for these interviews is published separately. There was a target of 100 host interviews with 20 in each of the five areas. A total of 77 host interviews were achieved with a breakdown of 22 in Edinburgh, 17 in Glasgow, 15 in the East Neuk of Fife, 12 in Skye, and 11 in Fort William.
Community actors and local businesses qualitative research
Community actors were recruited through direct approach via email and telephone. The types of actors interviewed included local councillors, community councils, local development trusts, residents' associations and other local representative groups. A total of 55 individual community actors were interviewed through a combination of individual and small group interviews, undertaken by telephone and face-to-face.
Local businesses were recruited through direct approach, i.e. through email, telephone and a walk-in direct approach. The types of local businesses interviewed included private landlords, residential and STLs letting agents, social and affordable landlords, self-caterers, cafés, restaurants, gift shops, domestic and STLs cleaning businesses, traditional guesthouses and B&B owners, chambers of commerce, tour operators, housing developers and other businesses involved in the construction sector, and a social care business. A total of 56 local business representatives were interviewed through a combination of individual and small group interview, undertaken by telephone and face-to-face. It should be noted that in the rural areas, many participants often had overlapping interests so the distinction specified between community and business interests was often blurred. The discussion guide has been published separately.
Qualitative research description
The purpose of qualitative research was to provide in-depth understanding into particular issues, by exploring experiences, characteristics or behaviours. As is typical, the sample sizes were smaller than those used for the quantitative surveys. Reporting of qualitative data has been done through analysis and description of opinion, rather than counts or measures. The following descriptions are used here in reporting qualitative findings:
- All – everyone participating in the research made this point
- Most – more than half of participants
- Many – more than some but less than most
- Some – less than half but more than three participants
- A few – two or three participants
- One/an individual – just one person.
In this research quotes are used to illustrate key points, and may be drawn from the four different types of participants – residents, hosts, community actors and local businesses. They may also be drawn from the five different areas and all quotes are annotated with the place and type of respondent, e.g. (East Neuk of Fife, community). The use of quotes by particular places or type of respondent does not infer any weight of response by that area, rather it serves to illustrate a point which may be made by many different participants, and in several areas.
Limitations of the research
The research has been designed around a case study approach and was mainly exploratory in nature, undertaken through qualitative methods. The qualitative research was combined with small quantitative surveys in each of the five case study areas.
The qualitative research approach allows for intensive exploration and description of key issues, thus allowing for insights into participants' views. It still allows the researcher to make reasonable judgements as to the prevalence of such views, although the nature of qualitative research (i.e. the number of interviews typically achieved, the free-form nature of responses, etc.) means that it is not possible to generalise the findings of the qualitative research in a statistical sense to that of the wider population. The qualitative approach allows us, given the number of interviews involved in this research and the recurring themes found, to summarize and develop general propositions and conclusions on the basis of these specific case studies. Whilst the numbers of qualitative interviews at a local level are obviously lower, they are still sufficient for the researchers to make judgements as to any notable characteristics of a particular area or distinctions between those areas.
In relation to the quantitative surveys, the target number of responses was achieved for the residents survey (197 responses) and the hosts survey (227 responses to the online survey). However due to the respective methodologies, we cannot calculate a response rate for either and so it is not appropriate to quote a level of statistical significance to these surveys. As such the survey results should be treated as indicative only. In the case of the resident survey, the sample is based on the selected wards of the case study areas only and not on the population of Scotland as a whole. Other than measures to attain a geographic spread within each area, no specific quotas were set based on the population of these case study areas and no weighting has been applied. The research does not therefore claim to be based on a random sample and nor were detailed quotas set to reflect the population breakdown (and thus reflect the characteristics of a random sample). This said, the profile information set out elsewhere in this report and appendices suggests that a broad range of respondents was achieved across a range of demographic and other descriptive respondent criteria.
The host survey was again conducted within the case study areas; in this case on a self-completion basis with invitations being issued by Airbnb and through different social media platforms. We are not therefore able to quote a response rate for this survey or to identify how the respondent profile compares to the profile of those invited to respond. However, the profile information provided illustrates that a broad range of host types participated in the survey.
The report comments on survey results at a local level where we believe these suggest a notable variation in views across areas. However, the limitations described above also apply at the local level. The small number of responses at a local level means that again statistical significance at area level is not appropriate.