Methodology

Grouping Towns into Clusters

K-Means clustering is a data driven clustering procedure which seeks to maximise differences between clusters, with cases being as similar as possible to others in any cluster and as dis-similar to cases in other clusters as possible. Unlike other procedures it does not start from a theoretical basis and established groupings to which cases are added, nor does it merely divide the range of values for cases into classes. K-means clustering is based upon numerical distance between cases when represented by scores on the variables. The number of clusters dictates the number of centroids and the distances are measured from these. The cases are grouped by minimising the distances between cases. The process is iterative and stops when no cases can be moved between groups.

The clustering procedure involved repeating the process for different numbers of potential clusters. A judgement was made between too much detail (too many groups of 1 locality) and too much generality. Five cluster groups were selected with only one of these having only 1 representative locality.

F values are calculated which are indicative of the contribution of each variable to the final analysis. Final centre clusters are used to define the clusters. Not all cases in a cluster share all characteristics of the cluster to the same degree. They are generally more like each other, however, than the cases in other clusters. No time series data is included in this version of USP, so it is unable at this point to differentiate towns on different trajectories.

Locality size

It is logical to expect that locality size, in terms of resident population, will impact on a locality's function and service provision. The clusters therefore have to be disaggregated in terms of the size of the towns. The table shows the size categories used, which provide compatibility with the Scottish Government's Urban-rural classification size bands. (3)

Combining the typology with size categories 13 different groupings (the typology) were established (i.e. there were 7 types in the typology which contained no town entries).

Inter-Relationship Model

The next level of Understanding Scottish Places was to apply an inter-relationship model. The inter-relationship model is framed by the Centre for Local Economic Strategies (CLES) research over the last ten years about the resilience of place. (5) The CLES resilience model explores the assets and relationships which places have between the public, commercial, and social sectors and how that shapes the functioning of their economies. The inter-relationship model explores this further by identifying the extent to which places are reliant or otherwise upon neighbouring localities for these assets and relationships.

In the Understanding Scottish Places (USP) project, we have developed a way of exploring this through developing a set of indicators. These indicators enable us to define the relative independence, interdependence, and dependence of the 479 towns in Scotland with a population of over 1000 people. We have called this our inter-relationship model.

An independent town will have a high number of public; commercial; and social economy assets in relation to its population. This will include GP surgeries and charities. It will have a diverse sector base in terms of jobs. Residents will travel short distances to work and study and the town will attract people from neighbouring towns to access its assets.

A dependent town will have a low number of public; commercial; and social economy assets in relation to its population. It will be reliant on singular sectors in terms of jobs. Residents will travel longer distances to work and study and the town will be reliant on neighbouring towns for assets and jobs.

An interdependent town will sit somewhere between independent and dependent towns. For some public, commercial and social economy assets it may have a high number in relation to its population and for others a low number. A balance of people will work and study in the town with others reliant on neighbouring towns.

Approach

Twelve indicators are used which explore the inter-relationships within and between towns in Scotland. These indicators were chosen on the basis of data being nationally available to populate them. The indicators portray three things:

• There are indicators which detail the number of certain assets in the town when compared to its resident or working age population;
• There are indicators which detail the diversity of the business and employment base in the town;
• There are indicators which detail the distance people resident in the town travel to work and to study.

To populate each indicator, we have gathered data from a range of sources for each of the 479 towns in Scotland with a population of over 1000 people. The 11 indicators and associated data sources are as follows:

• Number of registered charities - this is the number of charities based in the town in relation to the resident population. Data has been gathered from the Scottish Charity Regulator (OSCR);
• Number of GP’s and Dentists - this is the number of GP’s and Dentists in the town in relation to the resident population. Data has been gathered from Registers of GPs and Surgeries at the Information Services Division at NHS Scotland;
• Number of hospitals - this is the number of hospitals in the town in relation to the resident population. Data has been gathered from the Information Services Division at NHS Scotland;
• Number of children in primary schools - this is the number of children in primary schools based in the town in relation to its resident population. Data has been gathered from Education Scotland and the Scottish Council of Independent Schools;
• Number of children in secondary schools - this is the number of children in secondary schools based in the town in relation to its resident population. Data has been gathered from Education Scotland and the Scottish Council of Independent Schools;
• Number of jobs - this is the number of jobs in the town in relation to its working age population (16-64). Data has been gathered from the Business Register and Employment Survey;
• Diversity of jobs - this is number of jobs in particular sectors in the town and the extent to which it is diverse or reliant in sector terms. Data has been gathered from the Business Register and Employment Survey;
• Public sector jobs - this is the number of jobs in the town in the public sector in relation to all jobs. Both low and high numbers of jobs are a sign of reliance on either the public sector or other sectors. Data has been gathered from the Business Register and Employment Survey;
• Number of shops - this is the number of shops in the town in relation to its resident population. Data has been gathered from the Scottish Assessors Association;
• Distance travelled to work - this is the distance travelled by the working age residents of the town to reach their job. Data has been gathered from the 2011 Census;
• Distance travelled to study - this is the distance travelled by students resident in the town to reach the place of their studies. Data has been gathered from the 2011 Census;
• Diversity of retail offer [New to USP 2] - this is the percentage of retail types in the town in relation to 37 different retail types. The higher the percentage the greater diversity of retail types. Data has been gathered from the Business Register and Employment Survey.

The data has then been analysed utilising the appropriate one of three portrayals outlined earlier (number of assets; diversity of business and employment base; and distance travelled). For each indicator, towns have then been split into sevenths depending on their position across the 479 towns and given an appropriate score. Towns in the top seventh on each indicator (a high number of shops per resident, for example) have scored 6 and towns in the bottom seventh (a low number of shops per resident, for example) have scored -6 with increments of 4, 2, 0, -2, and -4 in between.

Once we have analysed each indicator for each town we have then added up the individual scores to derive a total for the town. The highest aggregated total score is 52 with the lowest -58. We have then taken the difference (110) between highest and lowest and split the towns into seven equal increments. For example, towns scoring between 36 and 52 are in the top increment. They have been subsequently been provided with the following assessments.

• First increment - Independent
• Second increment - Independent to Interdependent
• Third increment - Interdependent to Independent
• Fourth increment - Interdependent
• Fifth increment - Interdependent to Dependent
• Sixth increment - Dependent to Interdependent
• Seventh increment - Dependent

This gives each town their overarching score and assessment using our inter-relationship model.

Independent towns have a high number of assets in relation to their population; a strong diversity of jobs; and residents travel shorter distances to travel to work and study. These towns will attract people from neighbouring towns to access their assets and jobs.

Independent to Interdependent towns have a good number of assets in relation to their population. These towns have a good diversity of jobs; and residents on the whole travel shorter distances to travel to work and study. These towns attract people from neighbouring towns to access some of their assets and jobs.

Interdependent to Independent towns have a good number of assets in relation to their population. They have some diversity of jobs; and residents largely travel shorter distances to work and study, although some travel longer distances. These towns attract people from neighbouring towns to access some of their assets and jobs.

Interdependent towns have a medium number of assets in relation to their population; average diversity of jobs; and residents travel a mix of short and long distances to travel to work and study. These towns are attractors of people from neighbouring towns who come to access some assets and jobs but they are also reliant on neighbouring towns for other assets and jobs.

Interdependent to Dependent towns have a low number of assets in relation to their population. They have some diversity of jobs; and residents travel largely longer distances to work and study, although some travel shorter distances. They are reliant on neighbouring towns for some assets and jobs.

Dependent to Interdependent towns have a low number of assets in relation to their population. They have a poor diversity of jobs; and residents on the whole travel longer distances to work and study. They are reliant on neighbouring towns for some assets and jobs.

Dependent towns have a low number of assets in relation to their population; a reliance on one sector for jobs; and residents travel longer distances to work and study. They are reliant on neighbouring towns for assets and jobs.

Methodology Note on Change Indicators Introduced to USP 3

USP3 introduces measures which track changes to towns. This has involved measuring changes to the size of settlements, to jobs in the town and to sources of change such as migration. These came with their own data handling challenges. This methodological statement provides a brief description of the ways in which we defined and handled the requisite data. In Scotland the postcode is the building block for all other spatial data. Localities combine postcodes with any one or more of the following: population densities of greater than 5 people per hectare, more than 2.1 postal addresses per hectare more than 0.1 non-residential addresses per hectare.

Following the 2011 census the 2010 localities were re-examined and postcodes added or taken away reflecting changes to any of these densities. 2012 boundaries thus accounted for observed population change. So, where a locality has expanded or contracted spatially this is reflected in the new boundaries. The process also resulted in some other changes such as new localities with more than 1000 people, localities which fell below the 1000 population, splitting of localities and mergers of localities. At this point in the USP process it was agreed by the consortium that the original 479 towns would remain the baseline and so as far as possible these were re-constructed from 2012 data. Thus, no new localities have been included and split localities were joined back for the purposes of USP3. This was justified on the grounds that the 2011 census data was not recalculated for the 2012 boundaries making it impossible for the consortium to provide consistent data for the 2012 boundaries. GRO Scotland kindly calculated the population and household totals for the 2001 census based on 2010 boundaries giving a retrospective comparison. Going forward the 2010 boundaries with 2011 data can be compared with the new 2012 boundary data giving a reflection of the spread or contraction of different localities. Additionally, interim population data for 2016 has been included as applied to the 2012 boundaries. It should be noted that interim data is not the same as the census data as it is based on estimates arising from registers of births and deaths and medical registrations. They are estimates rather than counts.

Migration data is taken from the 2011 census at the original 2010 boundaries. The measure used in the net migration rate (inflow - outflow/100 population. The census definition of a migrant is based on change of address in the year up to the census date. Children less than 1 year old are excluded. In the case of USP migration was spatially defined as those moving in or out of the locality (be it from another locality, the wider local authority, the rest of Scotland, the rest of the UK or from abroad). Change of residence within the locality is not included.

Footnotes

1. National Records of Scotland (2014) Mid-2012 Population Estimates for Settlements and Localities in Scotland
2. Shepherd, J. (2009) A typology of the smaller rural towns of England , Birkbeck College, university of London.
3. Lupton, R., Fenton, A., Tunstall, R. and Harris, R. (2011) Place typologies and their policy applications , DCLG Case Report 65.
4. Scottish Government (2012) Urban-rural classification.
5. CLES (2011) Productive Local Economies: Creating resilient places and CLES (2014) Developing resilient town centres.

The USP typology is based purely on data from the Scottish Census 2011. The USP inter-relationship assessment also uses two indicators from the census. More information about census data can be found at www.scotlandscensus.gov.uk. ",Methodology,2015-05-18 13:08:44.000,"Understanding Scottish Places platform has been created using national data sets - a new typology of Scottish towns, and an assessment of towns’ inter-relationships.