The aim of this presentation is to investigate claims of decreased segregation yet also of ‘white flight’ from English cities during the period from 2001 to 2011. It does so supplementing a traditional measure of segregation, the dissimilarity index,
with measures comparing differences between adjoining small areas. Together these measures provide insight not only into the amount of segregation but also its spatial configuration within local authorities, including the degree to which different ethnic groups are clustered together of dispersed across the authorities. An analysis of change is then undertaken, asking whether the neighbouring small areas with greatest differences in their ethnic compositions in 2001 become more or less dissimilar by 2011, and whether those changes are caused by more population mixing or by the withdrawal of the White British population from those areas. Motion charts also are presented to warning against over-simplification and ‘one-size-fits-all’ explanations, stressing the individual trajectories of different local authorities.
General Principles of Intellectual Property: Concepts of Intellectual Proper...
Motion Charts, White Flight and Ethnic Cliffs?
1. Motion Charts, White Flight
and Ethnic Cliffs? Ethno
demographic change in the
2011 Census
Richard Harris
University of Bristol
2. Abstract
• The aim of this presentation is to investigate
claims of decreased segregation yet also of
‘white flight’ from English cities during the
period from 2001 to 2011.
• To supplement a traditional measure of
segregation, the dissimilarity index, with
measures comparing differences between
adjoining small areas.
• Look at whether the neighbouring small areas
with greatest differences become more or less
dissimilar by 2011, and how so.
• Motion charts also are presented stressing the
individual trajectories of different local authorities.
3. Overview
• Provide the context (the debate about ‘white
flight’).
• Argue that a decrease in a traditional segregation
index is not, in itself, evidence of decreasing
segregation.
• Provide an example of a spatial index (though it
does not resolve the above).
• Consider some different scenarios.
• Provide some qualitative evidence (not
necessarily reliable) to caution against dismissing
a ‘white flight’ partial explanation for demographic
change.
14. And yet…
• Ethnic segregation is falling in England (Catney,
2013)
• A lower proportion of the White British population
now living in majority White neighbourhoods than
they did in 2001 – a finding that is as true of
London alone as for the whole of England
(Johnston et al., 2013).
• “Academics cannot agree about whether [Britain]
is becoming more integrated or more segregated
in the wake of the unprecedented immigrant
inflows of the past 15 years” (Goodhart, 2013a,
p.47).
18. Measures of Segregation
• Typically,
• The index of dissimilarity,
I ∝ Oi − Eii=1
n
∑
D =
1
2
xi
X
−
yi
Yi=1
n
∑
X = xii=1
n
∑
Y = yii=1
n
∑
19. Criticisms of such indices
• Not easy to interpret over time
• It assumes a ‘closed system’ and can lead
to false conclusions:
Dt1 − Dt0 f Δ xi X[ ],Δ yi Y[ ]( )
t0 t1 Assume D1 < D0
Is it decreased seg.?
20. Criticisms of such indices
• Have no measure of statistical
significance
• But then what is the benchmark to
compare against? (Who expects
populations to be randomly distributed?)
21. Criticisms of such indices
• Segregation is usually regarded as an
outcome or cause of spatial processes of
separation.
• But the index is largely blind to
geography.
• All the following produce ‘total
segregation’ (D = 1)
22. Local Measure of Dissimilarity
• Take the parts
• Add a spatial weights matrix
• Creates a spatially smoothed D
D = d1 + d2 + d3 +...+ dn
d1 =
x1
X
−
y1
Y
, etc.
Di
*
∝ wijj=1
n
∑ dj
23. A slight alternative
• A weighted comparison of neighbours
– Local Index of Spatial Association (LISA)
• Replace:
• With:
• For example, O is the proportion of the
population White British and w is a
contiguity matrix…
Ii
*
∝ wij Oj − Ejj=1
n
∑
Ii
*
∝ wij Oi −Ojj=1
n−1
∑
25. A measure of discontinuity
• Example
– There is a census small area in Tower
Hamlets where 142 of its 258 residents
classified themselves as White British in the
2011 Census, and a further three as
Bangladeshi.
– Meanwhile, in a neighbouring area, 23 of the
211 residents are White British, and 142
Bangladeshi.
• The discontinuity can be calculated by
direct analogy to the dissimilarity index:
| 142/165 – 3/145 | = 0.840
26. Local Measure of (Max) Discontinuity
• In principle:
– For each small area, i, find the neighbour that
gives the maximum dissimilarity value, given a
minimum joint population threshold
X = xi + xj
Y = yi + yj
if X ≥ 30, Y ≥ 30 & wij =1,
dij = wij
xi
X
−
yi
Y
else
dij = NA
Dci
*
= max dij( )
Dc =
1
n
Dci
*
i=1
n
∑ ∀Dci
*
∉ NA{ }
28. The algorithm
1. Pairs each small area (i) to a neighbour (j) that
is within 100 metres of its border, choosing that
neighbour which gives the maximum value of di.
A table is created with each i and its initial
pairing, j.
2. Searches through the table to identify and to
remove those rows containing duplicated values
of j, retaining only the maximum cases.
3. For those areas (i) expunged by Step 2,
searches again for a new pair from a list of
unmatched js, still using maximum value and
neighbour criteria.
4. Cycles through Steps 2 and 3 until no further
pairings are possible.
29. A by-product
• To count the cases that are omitted by the
population threshold and express them as
a proportion of the total pairings. This will
give a simple measure of how much the
population groups cluster within the study
region: the greater the number, the more
often one of the population groups is
found in neither of the neighbouring small
areas (or in low numbers within them).
36. Task
• Identify those neighbours with the
greatest (Asian – White British)
discontinuity scores in 2001
• See if those differences have increased or
decreased to 2011
• Explanation for the change
39. What happened next?
• Of the 590 in 2001, 500 can be matched
to Census small areas in 2011 and, of
those, 71% have a decreased
discontinuity score.
• Of the 5490 in 2001, 5000 can be
matched and, of those, 74% have a
decreased score.
40. For the matched 500
• In 39% of cases
– The number of Asians residing in the area
with the lower proportion of Asians in 2001
had increased. The number in the neighbour
with the higher proportion of Asians had
decreased.
• In 49% of cases
– There are now more Asians living in both the
neighbours.
• In 69% of cases
– The number of White British in both areas had
decreased.
41. For the matched 5000
• In 32% of cases
– The number of Asians residing in the area
with the lower proportion of Asians in 2001
had increased. The number in the neighbour
with the higher proportion of Asians had
decreased.
• In 51% of cases
– There are now more Asians living in both the
neighbours.
• In 69% of cases
– The number of White British in both areas had
decreased.
42. • Finney and Simpson (2009) remind us, it
is a myth that minority groups want to live
in segregated neighbourhoods.
• More evidence…
46. But what about ‘White Flight’?
• Are apparent falls in segregation driven
more by the White British population
leaving areas rather than increased
mixing?
49. Wordle of the highest rated comments
on the Mark Easton article
50. In response to a Mark Easton article
• Why have the White British left London?
– “Some of the Areas have become very Asian
oriented with shops that cater more for Asian
people Halal Butchers, Bangladeshi Bakers,
Mosques and Temples, local shops that are
dominated by Asian foods and spices, and
that cater very little for non Asian tastes. Is it
any wonder the White people move out?
People feel like strangers in the town where
they were born.”
51. In response to the Mark Easton article
• Why have the White British left London?
– “They have left because they've been pushed
out by migrants. A few weeks ago we heard of
some Islamic "men" castigating a man for not
being Muslim, asking if he was gay, and telling
him to "get out, this is a Muslim area”. Their
exact words were played on the radio so there
was no doubt over it. Time for the govt. to do
something about such migrants”