2016 New Ward Boundaries Guidance on calculating statistics for the new 2016 wards

Transcription

1 2016 New Ward Boundaries Page 1 Performance, Information and Intelligence 2016 New Ward Boundaries Guidance on calculating statistics for the new 2016 wards Briefing Note Version 2 - March 2016

2 2016 New Ward Boundaries Page 2 Contents Quick Guide to Calculating Statistics for 2016 Ward Boundaries...3 Guidance on calculating statistics for the new 2016 wards Introduction Provision of statistics The issue Government Statistical Service Best-fit Geography Policy Calculating statistics for new 2016 statistical ward boundaries Terminology and Variable names Bristol City Council policy and planning boundaries Partner boundaries Availability of published statistics some examples Resources Further information... 10

3 2016 New Ward Boundaries Page 3 One Page Guide Calculating Statistics for 2016 Ward Boundaries In order to produce statistics that comply with Government Statistical Service (GSS) Geography Policy guidance, property/household/person level, or postcode level or Output Area level data should be used to aggregate up to 2016 wards. When calculating statistics for the 2016 wards, the following methods should be applied: 1. Property, Household and Person level data to 2016 wards For statistical purposes, data should be allocated to the 2011 Output Area within which the property or individual falls, then use the Output Area to 2016 Statistical Ward Lookup (OA11PWC-STATWD16). 2. Postcode level data to 2016 wards Use the National Statistics Postcode Lookup (NSPL). 3. Output Area level data to 2016 wards Use the Output Area to 2016 Statistical Ward Lookup (OA11PWC-STATWD16). Where statistics are not available by property/household/person level, or postcode level or Output Area level, then the best option is to wait until 2016 ward level statistics are officially published from 2017 onwards. Lower Layer Super Output Area level data When statistics are only available at LSOA11, in the first instance consider using the data at LSOA level and mapping it across the city with the 2016 ward boundaries shown on top. If however, 2016 ward level statistics are required from LSOA level data, then use the Lower Layer Super Output Area to 2016 Statistical Ward Lookup (LSOA11- STATWD16). Please note that any statistics calculated for 2016 wards using LSOA level data will be less accurate than aggregating individual/household, postcode or Output Area level data. Any 2016 ward level statistics aggregated up from LSOA11 data should be considered to be interim estimates, to be used until more accurate official 2016 ward statistics are published Ward level data This cannot be translated to 2016 wards. For non-statistical and administrative purposes more accurate mapping should be applied so that individual addresses are allocated to the exact ward within which they fall. For example when allocating electors to a ward or when using ward finder to tell people which ward they live in.

4 2016 New Ward Boundaries Page 4 Guidance on calculating statistics for the new 2016 wards 1. Introduction This guidance note will outline the implications of the new ward boundaries in terms of the provision of statistics, Bristol City Council policy and planning boundaries and partner boundaries. It will also outline the availability of published statistics and list resources available to assist with creating statistics for the new wards. 2. Provision of statistics 2.1 The issue New 2016 Electoral Wards were introduced in Bristol City Council in May The new ward boundaries do not fit to the boundaries for which statistics are published. In order to be able to produce statistics and profiles for the new 2016 wards, a solution needed to be found whereby a consistent approach to the calculation of statistics for the new wards was agreed. Performance, Information and Intelligence within Bristol City Council, has tested a number of approaches and assessed their advantages and disadvantages. The optimum solution chosen uses the same methodology adopted by the Office for National Statistics, which is based on a Best-fit allocation of Output Areas to the new wards. By adopting this approach, any statistics calculated by BCC will be consistent with any statistics produced by ONS, including any denominators that we may wish to use in order to calculate rates. 2.2 Government Statistical Service Best-fit Geography Policy ONS statistical geographies nest within each other Output Areas (OA11) nest within 2011 Lower Layer Super Output Areas (LSOA11s) and LSOA11 within 2011 Middle Layer Super Output Areas (MSOA11). Ward boundaries are an administrative geography and do not nest within this statistical geography hierarchy. Wards across the country did match the statistical geography hierarchy when it was introduced in 2004, but any ward boundary changes since mean that wards will no longer match. When calculating statistics, Government Statistical Service (GSS) Geography Policy is to use Output Areas as their building blocks for higher level geographies including statistical geographies such as LSOAs and MSOAs and administrative geographies such as wards. Point or postcode data is allocated to an OA and that whole OA is then allocated to a higher level geography based on its populationweighted centroid. As such if a new ward boundary cuts across an OA, then any statistics that ONS publish will be based on allocating that whole OA to the ward within which its population-weighted centroid falls.

5 2016 New Ward Boundaries Page 5 In Bristol, although our OA11 and LSOA11 fit to the 2011 wards, these statistical geographies will not nest within the new 2016 ward boundaries. Since statistical geographies remain largely unchanged over time, they will not fit to the new 2016 wards now or in future. There are a number of advantages to using the best-fit approach including: a simple methodology; clear allocations; one version of the truth for internal consistency especially when calculating rates (ie the numerator and denominator match); and no arbitrary splits of OAs and population characteristics. 2.3 Calculating statistics for new 2016 statistical ward boundaries i. Property, Household and Person level data BCC databases held at property, household or person level that include a ward reference will need to be allocated to 2016 statistical wards and 2016 geographical wards from May 2016 onwards. Assuming that all databases have a current UPRN, this should be a simple matching process. Business Systems automatically updated after the 5th May include: Uniform, Civica, Bristol Waste, Lagan, Enterprise System Bus (ESB), Liquid Logic and the Electoral Roll. ii. Postcode level data Figure 1: Postcodes, Output Areas and 2016 wards Postcode level data (ie full unit postcode BS1 5UY) should be aggregated to higher level geographies using the ONS Postcode Lookup (NSPL). The Postcode Lookup lists all postcodes and allocates them to an OA11. The OA11s are then allocated to

6 2016 New Ward Boundaries Page 6 higher geographies (including 2016 wards) based on the population weighted centroid (PWC) of the OA11. This ensures that any statistics used as numerators match the ONS denominators since both sets of data will be based on best-fit. The example in Figure 1 shows a number of postcodes (pink dots) within the same Output Area but straddles 3 different 2016 Wards. All four postcodes are allocated to 2016 Filwood ward based on the location of the Output Area s population weighted centroid (blue star). A local version of the NSPL has been created which includes the new 2016 statistical ward references. The actual 2016 geographical ward has also been included for reference. Only a very small proportion of postcodes are allocated differently between the 2016 statistical ward and the 2016 geographical ward. Based on postcodes as at November 2015, 97% of live postcodes are allocated to the same 2016 ward using the best-fit whole OA-based methodology as they would be if they were allocated to the ward based on the exact location of the postcode. iii. Output Area level data The smallest geography for which population and census statistics are available is by Output Area. OA11 statistics should be allocated to 2016 wards and other higher level geographies based on the population weighted centroid (PWC) of the OA11. In the example shown in Figure 2, the OA11 is split between three 2016 wards but would be allocated to 2016 Filwood ward based on the OA11 PWC (shown as a star). Figure 2: Output Areas and 2016 wards A lookup table is available. This ensures that any statistics used as numerators match the ONS denominators since both sets of data will be based on best-fit. Only

7 2016 New Ward Boundaries Page 7 a small proportion of OA11s are split by the new 2016 ward boundaries - 82 out of 1,368 (6%) of OA11s. iv. Lower Layer Super Output Area level data PLEASE NOTE THAT THE GUIDANCE FOR LSOA DATA HAS CHANGED FROM THE ORIGINAL GUIDANCE PROVIDED IN THE FEBRUARY 2016 VERSION OF THIS BRIEFING NOTE Where statistics are not available by postcode or OA11, then the best option is to wait until 2016 ward level statistics are officially published from 2017 onwards, as these will have been calculated by ONS using original raw data. When statistics are only available at LSOA11 level (eg Indices of Deprivation, Child Poverty), then consider just using the data at LSOA level and perhaps mapping it across the city with the 2016 ward boundaries shown on top. If however, 2016 ward level statistics are required from LSOA level data, then please use the method outlined below. Please note that any statistics calculated for 2016 wards using LSOA level data will be less accurate than aggregating either individual/household, postcode or Output Area level data. Any 2016 ward level statistics aggregated up from LSOA11 data should be considered to be interim estimates, to be used until more accurate official 2016 ward statistics are published. Due to the larger size of LSOAs, it is not practical to use the best-fit method for allocating LSOAs to new wards based on the population weighted centroid (PWC). For the LSOA11s that are split by the new ward boundaries, the proportion in each 2016 ward has been calculated using the OA11 population-weighted centroids to calculate the proportion of the LSOA11s population within each 2016 ward. Using this method, out of 263 LSOA11s in Bristol, 230 can be wholly allocated to a 2016 ward and 33 LSOA11s are given proportional splits to the new ward boundaries. In the example shown in Figure 3, Parson Street LSOA is split between three 2016 wards. Using the calculated proportions, Parson Street would be split between Bedminster ward (0.63), Bishopsworth ward (0.19) and Filwood ward (0.19). So, LSOA11 should be split according to the proportions provided in the new LSOA11-STATWD16 lookup table (this lookup table replaces the previously provided LSOA population-weighted centroid lookup table). By using the LSOA11 apportionment method, there is no need for an alternative 2016 ward level population denominator to that which is published. The apportionment is also a closer reflection of reality at ward level than the LSOA11 PWC best fit which is quite crude at this geographical level and a bad fit for some wards.

8 2016 New Ward Boundaries Page 8 Figure 3: Lower Layer Super Output Areas and 2016 wards v. Current Ward level data Many nationally published datasets are only available by frozen wards ie ward boundaries as at either 2001 or 2011 in order to enable comparisons of change over time. These statistics cannot be re-aggregated to 2016 wards. 2.4 Terminology and Variable names 2016 Statistical Wards and 2016 Geographical Wards For statistical purposes, where possible all data should be assigned to wards based on the GSS Best-fit methodology of Output Areas to higher geographies. Allocations made on this basis should be referred to as 2016 Statistical wards For non-statistical and administrative purposes more accurate mapping should be applied. For example, the electoral register electors should be allocated to the exact ward within which they live. Allocations made on this basis should be referred to as 2016 Geographical wards

9 2016 New Ward Boundaries Page 9 The following naming conventions should be used for variables: Variable name Geography OA11CD OA11NM LSOA11CD LSOA11LN WARD11CD WARD11NM GEWD16CD GEWD16NM STWD16CD STWD16NM STNP16CD STNP16NM 2011 Census Output Area ONS code 2011 Census Output Area local name 2011 Census Lower Layer Super Output Area ONS code 2011 Census Lower Layer Super Output Area local name 2011 Ward ONS code 2011 Ward name 2016 Geographical Ward ONS code 2016 Geographical Ward name 2016 Statistical Ward ONS code 2016 Statistical Ward name 2016 Statistical Neighbourhood Partnership code 2016 Statistical Neighbourhood Partnership name Note: The three St George wards should be ordered before other wards beginning with S. This is how the ONS codes are ordered. 3. Bristol City Council policy and planning boundaries Wards are used as a basis for a many policy and planning boundaries within BCC, including for example: Strategic Housing Market Area Zones; Community Infrastructure Levy charging and expenditure; CYPS Areas; Neighbourhood Partnerships. These boundaries should ideally be changed to reflect the new ward boundaries. 4. Partner boundaries A number of organisations and groups currently use ward boundaries, either individually or in aggregate, as a basis for their work. Notable examples include: Fire Service (Community Safety Sectors); Clinical Commissioning Groups; and Neighbourhood Policing Areas. 5. Availability of published statistics some examples The Office for National Statistics will only begin to publish statistics for the new 2016 wards from 2017 onwards. ONS will publish Mid-2016 Small Area Population Estimates for the first time for new 2016 wards based on the GSS best-fit methodology at the end of In the meantime, population estimates for new 2016 wards in Bristol have been calculated by Performance, Information and Intelligence using the best-fit methodology.

10 2016 New Ward Boundaries Page 10 The majority of 2011 Census statistics are available by OA11 and so can be reaggregated to 2016 Statistical Wards using the best-fit methodology. Some of the more detailed tables which are only available at ward level and above, however, will be not be available for 2016 wards Census data have been calculated for the new 2016 wards by Performance, Information and Intelligence using the best-fit methodology Ward Profiles are available here The profiles include Population, Quality of Life, Crime, Education and 2011 Census data for the new wards. The 2011 Census results together with some other ONS small area data sets are available from NOMIS A user-defined geography for the 2016 Statistical Wards has been created by Bristol City Council which is publically available to use on NOMIS. In the list of user-defined geographies select Bristol City Council (1499) and find STATWD16. Since the definitions are based on aggregations of OA11s, the 2016 Statistical Wards can only be used to extract statistics which are originally available at OA11 level. The 2015 Indices of Deprivation is only available by LSOA11 and should continue to be used at an LSOA level. When mapping the 2015 Indices of Deprivation, the new ward boundaries can simply be overlaid on top. DWP benefit statistics are currently only published for wards as in 2001 and LSOA01. As such it is not possible to re-aggregate to the new ward boundaries. 6. Resources The following lookup files are available on the Bristol City Council shared directory G:\BCC-Common\Ward 2016: Output Area to 2016 Statistical Ward Lookup (OA11PWC-STATWD16) Lower Layer Super Output Area to 2016 Statistical Ward Lookup (LSOA11-STATWD16) National Statistics Postcode Lookup (NSPL) for statistical geographies National Statistics Postcode Directory (NSPD) for geographical allocations Explanation of the difference between the National Statistics Postcode Lookup (NSPL) and the National Statistics Postcode Directory (ONSPD) These resources can be shared with non-bcc partners and other on request. 7. Further information Jayne Mills, Performance, Information and Intelligence/Strategic City Planning E: jayne.mills@bristol.gov.uk GIS Support team E: GIS_SupportTeam@bristol.gov.uk