Slideshare uses cookies to improve functionality and performance, and to provide you with relevant advertising. If you continue browsing the site, you agree to the use of cookies on this website. See our User Agreement and Privacy Policy.

Slideshare uses cookies to improve functionality and performance, and to provide you with relevant advertising. If you continue browsing the site, you agree to the use of cookies on this website. See our Privacy Policy and User Agreement for details.

Successfully reported this slideshow.

Like this presentation? Why not share!

- Motion Charts, White Flight and Eth... by Rich Harris 8720 views
- White flight, ethnic cliffs and oth... by Rich Harris 8469 views
- Count on us? A crisis of numeracy i... by Rich Harris 2214 views
- Actively Waiting in Advent by Rich Harris 418 views
- An introduction to Mapping and Spat... by Rich Harris 11327 views
- An Unequal Society: what must Chris... by Rich Harris 290 views

8,716 views

Published on

No Downloads

Total views

8,716

On SlideShare

0

From Embeds

0

Number of Embeds

7,872

Shares

0

Downloads

0

Comments

0

Likes

1

No embeds

No notes for slide

- 1. + Optimal models of segregation Richard Harris
- 2. + Classic segregation index I = k obsi −expii∑ D = 0.5 ai aii ∑ − bi bii ∑i∑
- 3. + Classic segregation index n Literally the sum of its parts n And those parts are places (locations within the overall study region) n Which means they can be georeferenced i (xi, yi)
- 4. + Classic segregation index n The georeference is superfluous n Only need to know that i is within the study region n Really just the average difference between two variables n Hence…
- 5. + Classic segregation index n … all these have the same index value (D = 1) n “the checkerboard problem”
- 6. + Same lived experience of residential segregation?
- 7. + Five dimensions of residential segregation (Massey & Denton, 1988) n (Un-)Evenness n which is what the classic segregation index measures n Exposure n Probabilistic measures n Concentration n Centralization n Clustering n Concerns the “checkerboard problem”
- 8. + Clustering n Jakubs (1981) and Morgan (1983) propose distance-based measures based on linear optimization. n Recall, for example, n For each location (i) determining whether obs > exp or obs < exp and therefore whether it has a surplus or deficit of the target population. I = k obsi −expii∑
- 9. + The optimization n Minimise the distance the target population would have to travel between (between the georeferences, i.e. centroids) to achieve an even distribution. n Both Jakubs and Morgan then calculate Dmin / D* where d* is the minimum distance under some hypothetical condition of maximum segregation.
- 10. + A problem n “After some initial experimentation … we quickly discovered that computing costs and machine requirements quickly became prohibitive for large urban areas” n Massey and Denton (1988, p.296) n But that was then… n … well before the R optimization package lpSolve became available (for free)
- 11. + Some tweaks n I am not convinced by the need to construct a ratio (i.e. Dmin / D*) n Instead, use Dmin / Σa – the average distance a person in group a would have to travel to create evenness (with group b) n Dmin is a function of the distances between zone centroids.Try: n Use nth nearest neighbour instead of Euclidean distance and minimise that instead n (Dmin / Σa) / meandc where meandc is the mean inter-centroid distance
- 12. + Some Pearson correlations D Dsp (nn) Dsp (Euc) N D 0.415 0.763 0.210 Dsp (nn) 0.356 0.788 Dsp (Euc) -0.045 D is the standard index of dissimilarity (Asian Vs White British for the 75 authorities with greater than 5% population Asian*) Dsp (nn) is the Spatial Dissimilarity index using nearest neighbour distance. Dsp uses Euclidean distances (but divides by mean inter- centroid distances) N is the number of Output Areas in the authority * Asian here defined as Pakistani + Indian + Bangladeshi (largest groups)
- 13. + Blackburn with Darwen
- 14. + However… n This is not actually a measure of clustering n It conflates clustering and centralisation n Pearson correlation (Dsp, Distance between the mean centre for the entire population per local authority and the mean centre for the Asian population): 0.529
- 15. + Three choices n Decide what actually it is n Geographic measure of the residential separation of the White British from the Asian population with the local authorities n So, if you believe distance effects contact, it is a measure of exposure n Create a ratio against some hypothetical notion of ‘total segregation’ for the given geography n Control for the centralisation effects (e.g. using regression) …
- 16. P WBloss D Dsp Cl Wycombe Wolverhampton Wokingham Woking Windsor and Maidenhead Westminster Wellingborough Watford Warwick Wandsworth Waltham Forest Walsall Trafford Tower Hamlets Three Rivers Tameside Sutton Stoke-on-Trent Spelthorne South Bucks Solihull Slough Sheffield Sandwell Rochdale Redbridge Reading Preston Peterborough Pendle Oxford Oldham Oadby and Wigston Nuneaton and Bedworth Nottingham Newham Newcastle upon Tyne Milton Keynes Middlesbrough Merton Manchester Luton Leicester Leeds Kirklees Kingston upon Thames Hyndburn Hounslow Hillingdon Harrow Hackney Gravesham Enfield East Staffordshire Ealing Dudley Derby Croydon Crawley Coventry Charnwood Camden Calderdale Bury Burnley Brent Bradford Bolton Blackburn with Darwen Blaby Birmingham Bedford Barnet Barking and Dagenham Wycombe Wolverhampton Wokingham Woking Windsor and Maidenhead Westminster Wellingborough Watford Warwick Wandsworth Waltham Forest Walsall Trafford Tower Hamlets Three Rivers Tameside Sutton Stoke-on-Trent Spelthorne South Bucks Solihull Slough Sheffield Sandwell Rochdale Redbridge Reading Preston Peterborough Pendle Oxford Oldham Oadby and Wigston Nuneaton and Bedworth Nottingham Newham Newcastle upon Tyne Milton Keynes Middlesbrough Merton Manchester Luton Leicester Leeds Kirklees Kingston upon Thames Hyndburn Hounslow Hillingdon Harrow Hackney Gravesham Enfield East Staffordshire Ealing Dudley Derby Croydon Crawley Coventry Charnwood Camden Calderdale Bury Burnley Brent Bradford Bolton Blackburn with Darwen Blaby Birmingham Bedford Barnet Barking and Dagenham
- 17. P WBloss D Dsp Cl Wycombe Wolverhampton Wokingham Woking Windsor and Maidenhead Westminster Wellingborough Watford Warwick Wandsworth Waltham Forest Walsall Trafford Tower Hamlets Three Rivers Tameside Sutton Stoke-on-Trent Spelthorne South Bucks Solihull Slough Sheffield Sandwell Rochdale Redbridge Reading Preston Peterborough Pendle Oxford Oldham Oadby and Wigston Nuneaton and Bedworth Nottingham Newham Newcastle upon Tyne Milton Keynes Middlesbrough Merton Manchester Luton Leicester Leeds Kirklees Kingston upon Thames Hyndburn Hounslow Hillingdon Harrow Hackney Gravesham Enfield East Staffordshire Ealing Dudley Derby Croydon Crawley Coventry Charnwood Camden Calderdale Bury Burnley Brent Bradford Bolton Blackburn with Darwen Blaby Birmingham Bedford Barnet Barking and Dagenham Wycombe Wolverhampton Wokingham Woking Windsor and Maidenhead Westminster Wellingborough Watford Warwick Wandsworth Waltham Forest Walsall Trafford Tower Hamlets Three Rivers Tameside Sutton Stoke-on-Trent Spelthorne South Bucks Solihull Slough Sheffield Sandwell Rochdale Redbridge Reading Preston Peterborough Pendle Oxford Oldham Oadby and Wigston Nuneaton and Bedworth Nottingham Newham Newcastle upon Tyne Milton Keynes Middlesbrough Merton Manchester Luton Leicester Leeds Kirklees Kingston upon Thames Hyndburn Hounslow Hillingdon Harrow Hackney Gravesham Enfield East Staffordshire Ealing Dudley Derby Croydon Crawley Coventry Charnwood Camden Calderdale Bury Burnley Brent Bradford Bolton Blackburn with Darwen Blaby Birmingham Bedford Barnet Barking and Dagenham
- 18. + Closing Thoughts n There is no longer a technological barrier for treating the measurement of segregation as a (spatial) optimization problem. n But it is not a measure of clustering per se n It still takes a while to calculate n And the ‘most segregated’ [sic] authority based on the principal component loadings?
- 19. + Newham n It is because 36% of Newham’s residential population is Asian and it also lost 37% of its White British population from 2001 – 2011 n The next is Redbridge (33% Asian; lost 30%) n Then Blackburn (unevenly distributed and spatially clustered Asian population) Prop. Asian 0.05 to <0.15 0.15 to <0.30 0.30 to <0.60 0.60 to 0.81 Omitted

No public clipboards found for this slide

×
### Save the most important slides with Clipping

Clipping is a handy way to collect and organize the most important slides from a presentation. You can keep your great finds in clipboards organized around topics.

Be the first to comment