SASI Home| Publications | Data | Maps | SASI Staff
Dorling, D. (1991) The Visualization of Spatial Structure, PhD Thesis, Department of Geography, University of Newcastle upon Tyne

Chapter 5: Transforming the Mosaic

Yet within any town as in the region as a whole there is a pattern. The poor housing, schools and levels of unemployment will tend to be concentrated in certain districts — as they are concentrated in inner city areas of the large conurbations of this country. At the level of the region, too, there is a pattern, increasingly clear and changing.
[SEEDS 1987 p.6]

5.1 Still Images of Change

Counts, measures, votes and all the other figures we use to build a picture of our social landscape are collected regularly because it is recognised that the picture changes. People’s positions and aspirations alter. Compared to the static picture, much less research has been addressed to looking at change. This may be partly the result of the change being generally slow, but much is due to the difficulties of displaying change69. These difficulties range from simple problems of gathering the information (temporal discontinuities), to the technical difficulties of processing it (what type of change is to be seen), to finally, the imaginative hurdles that have to be crossed in portraying it (creating still images of change).

Change can be looked at in different ways; absolute change (of so many people), relative change (of a certain percent), or relative expected deviation (higher or lower than the average change). The first two measures give an indication of what is happening to the country as a whole, the third removes that factor and allows concentration on variations from place to place (Rhind D., Evans I.S. & Dewdney J.C. 1977, Norris P. & Mounsey H.M. 1983, Brusegard D. & Menger G. 1989, McKee C. 1989c, Savage M. 1989, O.P.C.S. 1991).

Change is important. Change shows how the landscape came to be, where it is going, and how much it can alter. It is all too easy to think that so strong a spatial structure as we know exists today will prevail tomorrow. Change can show the spread of division, the concentration of inequality and the instability of a situation.

5.2 Forming the Structure

In past research, geography often stopped at the first hurdle when gathering information about change; coping with temporal discontinuities. Temporal discontinuities occur when units of population have their spatial boundaries altered. Temporal discontinuity is continuously re-occurring, and is itself one aspect of change70. As people move, so do the collating boundaries move around them, eternally attempting to encompass them adequately . We need to encompass these changes within our pictures (see Appendix B).

Practically every person in Britain is counted in the census every ten years. The simplest single number to be gathered from this is the total change in population, an increase of however many thousand between 1971 and 1981. How is this simple loss or gain of people distributed over our landscape? During this period Britain undertook its greatest ever redistribution of administrative boundaries — everything altered. Very few figures collected before the mid 1970s could be directly compared with those that came after (Massey D.S. & Stephan G.E. 1977, Rees P.H. 1977, Craig J. 1988, McKee C. 1989a, 1989b, 1989d).

Geographers often addressed the problem of change with the crude solution of aggregation71. The method is to find a set of large areas either side of the time period whose summed figures can be directly compared. This solution causes a great deal of information to be lost; local patterns can no longer be seen, large areas arbitrarily appear uniformly good or bad when the more truthful picture is very different (see Print LXVII).

To see national patterns or regional or city-sized processes it is better not to use national, regional or city-sized spatial units. Rather, show the eye the finely detailed picture, and let the mind decide how much pattern does or does not exist. Only then can the decision be made whether to smooth the picture further . There are also many means other than indiscriminate geographical amalgamation which can be used to generalize an image72.

How, though, do we create these fine images of local change from two sources based upon small, but differing, areas? What I do is to recognise that any change in boundaries has only a very local effect (Figure 14). People are moved from one side of the line to the other. There is no need to abolish the line, simply to realize that a few people have been moved. A detailed image, where nothing but the boundaries has really changed, will simply appear a constant, slightly speckled shade. The eye interprets the fine dithering that will have been created by misplacement as a colour, not a pattern. The problem has been reduced away.

It is not always as simple as this. Between 1971 and 1981 we were fortunate that the total population, and its internal distribution, changed little. Advantage can be taken of this fact. For instance, the underlying cartogram which was used to portray it did not perceptibly change its shape. Over long periods, which are examined later in this chapter, the underlying base map changes with time. Secondly, most spatial units, of any size, will usually correspond to just one or two predecessors. Occasionally the relationships will be more complicated (Figure 15).

5.3 Structure Transformed

The most basic changes of population have been simply painted by making each block white where population fell, and black where it increased. The white holes of the major conurbations can be easily distinguished, as can the black rings of built up areas around them. Importantly, the truth of this generalization can be ascertained from just how clearly this pattern stands out73.

There are no woodlands and fields on the population cartogram, just the people-lands of inner cities, suburbs, small towns and villages — all in proportion to their populations. Some of these have been growing and some declining. Everywhere there has been great variation, from street to street, suburb to suburb. As I progressively smooth the image, averaging each cell of one hundred people with eight hundred of their neighbours, then two thousand, five thousand, eight thousand... we see a more and more generalized image of the process of population redistribution. Information that is perhaps more clear, but less real.

The changing distributions of the sexes and ages can similarly be depicted (see Print VII). For the distribution of the sexes to alter, people must be born, die or move. Age obviously changes continuously with time, as well as irregularly over space (as people move). These two attributes are, however, interrelated, for as people age, men die earlier, and so changes in the proportions of the elderly are reflected by changes in the geography of the sexes. Similarly, more children will be born and brought up where there are more women. Again one influences the other. We could struggle to see these influences on three separate maps, one of the elderly, one of gender and one of children. How much better to show these interrelated changes in a single image by three colour shading. Now, though, rates of change rather than proportions of people are under scrutiny. What kind of change should we show?

Change is primarily a product of the time over which it is measured. There are no coastlines in time to delimit a period neatly, as we can a space. Using the state census, we can claim to have no choice but to use decennial intervals with no intervening information. How to cope with more fruitful situations is the subject of much of the rest of this thesis. For now, we only have two measures in time, and thus a single change to portray (Figure 16).

Aspects of the changes discussed above can themselves be examined more closely. Change in the spatial distribution of children, of different age groups, is examined. This image, which presents such a jumbled picture, tells us that there has been little uniform progression in these five year bands over time. The confusion is caused by families moving; very few remain in the same block for ten years. Unravelling the effects of migration is difficult. The next chapter examines it in detail. For now, note that even pictures which show no structure are showing something. Until you look at a picture you can only guess what you may or may not be able to see.

The last chapter introduced the major theme of migration to this thesis by looking at where people in Britain were born. Here we look at how those pictures are changing (Print LXXXVII). This could be done by seeing where those migrants have moved to, or from, but here we show how the proportion of lifetime migrants has altered in areas over the ten years between censuses. High levels of colour in the image indicates that either the proportion of lifetime migrants in that area has risen, or has not declined as much as elsewhere. Of those born in Britain, what is most striking is the return of the Welsh to the Valleys (or their unwillingness to leave them). Then there is the infusion of English born into the rest of Wales and highland Scotland74. The decline of all three national birthplace groups generally shows areas where immigrants have been moving in.

The picture of change in the proportions of those born in (all) Ireland, Asia and Africa depicts some interesting features (Print LXXXVIII). The rings of movement of lifetime international migrants out of the centre of London are distinctive75. This group of people had a major impact upon the changing social landscape of Britain in the 1970s as their moving influenced the basic age and sex distributions. They can also be seen to be moving to escape or seize the changes of opportunity that occurred (Craig J. 1980a, 1985b, 1987c, Mounsey H. & Clarke J.I. 1981b, Mounsey H.M. 1982a, 1982b, Peach C. 1982, 1984, Lawton R. 1986, Champion A.G. & Congdon P. 1988).

5.4 Variable Employment

One of the most variable attributes of our social landscape is employment, or the lack of it. The obvious extension of the above methods has been performed to illustrate the transformation of people’s lives from 1971 to 1981 which depended on where they live (Prints LXXXIX). The West Midlands increased its share of the unemployed and inactive by 1981; more people were working around London, and so on. These images show strong patterns, but just as indistinct images are not worthless, simple ones are not necessarily true.

Employment is a feature of our landscape that changes seasonally, the vegetation cover if you like. The single change over ten years hides great swings in the fortunes of places between those dates. Unemployment has been measured for areas, the size of towns, for every month since 1978, a complex spacetime series. To show the changes most simply we can paint a small image for each year, showing the deviation in each area from expected levels for that place and time. Such a series shows us how the spacetime pattern of unemployment deviates from what we would imagine, given a simple graph of time, and single cartogram of space (Prints XC & XCI).

The series of cartograms has areas shaded dark to indicate higher than expected levels of unemployment, moving towards white for lower than expected levels. At the end of the 1970s a Celtic fringe of high unemployment is apparent; by the end of the 1980s a very distinctive ring of low unemployment has grown around London76. In between are shown bad times for the people of the South, when unemployment was high everywhere and thus they too were doing relatively badly. The shading of these areas is as dependant on the limits of the time period as it is on the spatial limits of Britain. What is more, only a few years can be shown on a page (although at least years are a sensible amalgamation of months for counting unemployment). However, the beginnings of a picture of space and time is emerging (Frost M.E. & Spence N.A. 1983, Green A.E. 1983, Pinch S. & Williams A. 1983, Buck N., Gordon I., Young K., Ermish J. & Mills L. 1986, Dunn R. 1987b, Balls E. 1991).

The geology of industry changes much more slowly than that of employment, even though the latter follows changes in the former. Detailed information on people working in industry in many places has only been available towards the end of the 1980s and only then for the years that the census of employment has covered. The change for wards is shown between the 1984 and 1987 censuses, and an alternative colour scheme is used to depict which industries grew the most and which showed most decline (see Prints XXIV & XXV). The distribution of occupations, as might be expected, hardly altered at all over the ten years to 1981 (Print XCII)

5.5 House Price Inflation

A large sample of house sale information has been collected for the years 1983 to 1989 inclusive and has been converted for processing at the ward level. The static picture has been shown before (see Print LXXX), now inflation at every place in every year is shown. Here, though, a different shading scale from that used above is adopted. We wish to compare years as well as areas, so a fixed continuous shading is employed, light to indicate rapidly increasing prices, dark for slightly falling ones.

The picture is at first murky, a problem, perhaps, with the somewhat unreliable figures in the first year (Print XCIII). With so few and such different houses being sold by particular building societies in particular wards in particular years, spurious changes can be found. The mix of houses being sold in a ward is weighted by the national mix before a housing price can be calculated for the area. One outlying spot in ten thousand can easily occur. Far more notice should be taken when a clump is seen, for this is very unlikely to have arisen at random. Spatial and temporal smoothing could have been employed here, but it is remarkable how well the structure of local housing can be seen from the pictures of raw information.

Soon after 1983, a complex pattern begins to emerge of high inflation in the Home Counties and London, slowly moving out in a rough ring (Prints XCIV, XCV & XCVI). This picture mirrors the changes in unemployment described above. The increases become greater and greater, but a dark core begins to form in the centre (Print XCVII). Suddenly the darkness envelopes and the house price slump of 1989 is upon us77 (Print XCVIII). Only images could show how this began, preserving the detail, rather than averaging (Saunders P. 1989, Merrett S. & Sharp C. 1991, Coombes M.G., Champion A.G. & Monro M. 1991).

Still, when it has all finished, little has really changed. The static pattern for 1989 looks much like 1983, although the prices have trebled (Print XCIX). The images should be borne in mind, however, in connection with other changes, and also in connection with how they relate to changes about to be shown. Unemployment and inflation are claimed to be the major preoccupations of those who are about to vote.

5.6 Reshaping Votes

In considering voting in general elections, I have gone back in time before the era covered by most of this work to see how the changes of today compare with other years (see Appendix C). Ten general elections have been held since 1955 (by-elections have not been included, but they have never altered the government in power). What we are interested in are changes which occur at elections, which change governments and, as the political landscape always slowly changes, we want to be in a position to guess where it is going (Cox K.R. 1968, Gudgin G. & Taylor P.J. 1973, Miller W.L., Raab G. & Britto K. 1974, Busteed M.A. 1975, Crewe I., Särlvik B. & Alt J. 1977, Burnham W.D. 1978, Taylor P.J. 1982, Urwin D.W. 1982, Hurst M. 1986).

Images can be produced, showing which parliamentary constituencies changed hands between each contest (see Prints CXXXVIII & CXXXIX). These are most important in showing the geography of political success. Only those seats which changed hands are coloured, on the outside of their symbols by the former party, inside by the new holders. Between three parties there are six possible colourings, (between six different parties there are thirty). Here, the results of those changes, and something of what lies behind them, can be seen (Prints C, CI, & CII)

Showing the changing proportion of votes is more problematic. The British electoral system is dominated by three major parties. It is the swings between these which are of interest. The swing between three choices is a two-dimensional object (just as the simple swing between two choices, a basic change, is one-dimensional). A two-dimensional change can be shown in various ways. These pictures (see Prints CXLV & CXLVI) use arrows, the direction of which indicates the direction of change and the arrow length indicates magnitude of change. Pointing up was movement towards the Liberal party, left towards Labour, and right towards the Conservatives. The advantage of the parties having conventional directions (left, right and centre) as well as colours (red, blue and yellow), has been fully exploited. The arrows could be coloured by the existing shares of the vote, to show what political complexion the change was from.

What can be seen from these images of electoral fortune are uniform swings around the country with distinct variations depending on both geographical and political position78. The divergence in political behaviour grew stronger towards the more recent elections, as the arrows in different parts of the country began to head in different directions, taking the voting compositions of the seats and the pattern of victories with them. The dramatically changing fortunes of the liberal party are clearly shown (Denver D. 1989, Martis K.C. (ed.) 1989, Mishler W., Hoskin M. & Fitzgerald R. 1989, Denver D. 1990, Pattie C. 1990, Johnston R.J. & Pattie C.J. 1991, Sanders D. 1991).

It may not have been realised that these images encompass two geographical redistributions of constituency boundaries, both increasing the number of seats being contested. These changes are incorporated in the graphics, the cartograms for which were based on the electorates of their respective election. New seats squeeze in between their neighbours and old ones are squeezed out79. All the time the general shape of the country is changing as people migrate both out of cities, and, in general, to the South. The changing pattern of turnout is also interesting (Print CIII).

Politics is the animal life of our landscape. The political life can be very variable, but is less affected by season than is the vegetation cover of employment. Every aspect of our social landscape is subtly connected to everything else, as are all the changes and pictures and explanations. That is why it is worth looking at so many facets of life, rather than concentrating on illuminating one corner of one image in the gallery of pictures that exist to be seen.

5.7 Erosion and Deposition

The British social landscape changes. The industrial geology is most intransigent, but when it is altered all else must change. The population structure is stable, but reflects movement and the changing fashion for having children, the end of an older picture and the beginnings of a younger one. This is the soil of our landscape and it comes in many colours. The changing positions of lifetime migrants show several aspects of this restructuring through the movement of people, to where they choose to live and to where they are constrained to reside. The living parts of our landscape, those which change day-to-day, change fastest. The distribution of jobs, wealth, housing and how people vote depend upon all the other layers in the overall picture and on each other to an extraordinary extent80.

For key contributions to the debate on the geography of polarization in Britain see: Taylor P.J. 1979, Spence N.A., Gillespie A., Goddard J., Kennett S., Pinch S. & Williams A. 1982, Goddard J.B. & Champion A.G. 1983, Champion A.G., Green A.E., Owen D.W., Ellin D.J. & Coombes M.G. 1987, Hamnett C. 1986, 1987, 1989a, Townsend P. with Corrigan P. & Kowarzik U. 1987, Curtice J. 1988, Johnston R.J. & Pattie C.J. 1989a, 1989b, Halsey A.H. 1989, Pond C. 1989a.

How the landscape is changed depends on other forces which we have not yet examined. It is the flows of people, like flows of water, which both maintain and alter the picture. Every day the flow of people to work links industry to population. Over the days, months and years, people suddenly move in a quite different way: they move to different homes — they migrate. It is the streams of migration which sculpture our landscape, transporting its structure, depositing a new workforce and eroding the old. These flows of people, which maintain, and change the social landscape are the subject of the next chapter of this dissertation.

LXXXVII The changing distribution of British born place of birth, 1971-1981.
LXXXVIII The changing distribution of overseas bom place of birth, 1971-1981 (Colour).
LXXXIX The changing distribution of employment in Britain, 1971-1981 (Colour).
XC The space/time trend of unemployment in Britain by office areas, 1978-1990.
XCI The space/time trend of unemployment in Britain by counties, 1978-1990.
XCII The changing distribution of occupation in Britain, 1971-1981 (Colour).
XCIII The distribution of housing price inflation in Britain, 1983/1984.
XCIV The distribution of housing price inflation in Britain, 1984/1985.
XCV The distribution of housing price inflation in Britain, 1985/1986.
XCVI The distribution of housing price inflation in Britain, 1986/1987.
XCVII The distribution of housing price inflation in Britain, 1987/1988.
XCVIII The distribution of housing price inflation in Britain, 1988/1989.
XCIX The distribution of housing price in Britain, 1989.
C Voting composition by constituency, 1955-1987 (Colour).
CI The distributions of first placed party, 1955-1987 (Colour).
CII The distributions of second placed party, 1955-1987 (Colour).
CIII The distributions of non-voting by constituency, 1955-1987.


The 1971 and 1981 census geographies were linked at the enumeration district (ED) level. The majority of ED boundaries had not changed or were nearly identical, but in some places substantial alterations had occurred, for instance: where a new town had been built or old estate pulled down. The use of the "census tracts" designed by OPCS has been found to be far from adequate by McKee (1989). An alternative, far more flexible solution was devised. Only enumeration district centroids were known for each set of roughly 130,000 points. Two 2D tree data structures were built and the closest 1981 district to each 1971 found, and visa versa. Thus every ED in each set was connected to at least one in the other - but could be connected to any number, if required. Counts could then be compared.

Figure 14: Linking the Censuses

The following tables show how often a one to one link is achieved, and how it is unnecessary to combine up to 98 EDs in places (as is done with census tracts), when this approach is taken.

Links Between 1971 and 1981 Enumeration Districts

1971 EDs to each in 1981 1981 EDs to each in 1971





Total 1981 EDs   129211 Total 1971 EDs   125476

Note: 216 1971 enumeration districts had identical grid references of 600km East by 400km North and thus, lying in the the North Sea, were not linked to the 1981 set. These were probably used to record the population aboard ship.

Figure 15: How Closely Connected?

Observed change (O) can be measured in many ways between two times (T) and many places (i). For instance


Expected change can be calculated by:

Then deviation from expected is derived as:

Given six categories of housing and the national average prices (P) and proportions (W) of these, it is possible to estimate from the local distribution of prices (p), the average housing price (h), as either an arithmetic or geometric mean:

Figure 16: Measuring the Changes

69 [a] It can be claimed that in abstract terms:
Our maps are in one sense diagrams of geographic systems and their evolution. Many of them are — or were — cartographically communicated theories about global or regional geographic systems of resources and settlement. Many time series of maps are in one sense statements of theories, in cartographic language, about geographic development processes, about the functioning and the past and future evolution of some global or regional geographic system. Interpretations of the map patterns involve logical interpolation or extrapolation from mapped observations, in both space and time. Distinctively geographic models are also cartographic generalizations. As four-dimensional descriptions of the geographic evolution of resource and settlement systems, time series of maps are a fundamental element of geographical explanation. [Borchert J.R. 1987 p.388]

[b] But there is a paradox inherent in trying to show change between censuses:
Change in the context of the census occurs from one census to another through the necessity to illustrate the conditions present at the time of the census; therefore there is a distinct conflict between maintaining consistency from one census to another and the attempt to mirror relevant current conditions. The paradox is that the greater the real change which occurs between the censuses, then the greater the change that is required in terms of census questions, variable definitions, classifications, and geographical bases in order to reflect the changed conditions and hence the greater the difficulty in comparing the censuses in order to measure such change. [McKee C. 1989 p.3]

70 [a] Colour has been used, in this dissertation to show the time dimension in an illustration of population change between 1961 and 1991 (see Print CLV):
Perhaps the greatest potential is the time dimension. Maps are static views, and require corresponding approaches to data collection and map production. Electronic communication allows instantaneous distribution from one, continuously updated authority. The implications of this technical change extend not only to the way the data are viewed, but also to the way they are collected, and the infrastructure which has grown up around data collection. The decennial census, for example, was a solution to the problems of data collection and dissemination of the mid-19th century. Yet we have only the vaguest notions of how to exploit the new technology in this area; our perceptual systems are so geared to conventional display that we find it difficult, for example, to conceive of color being used to show the time dimension, or of how to structure and store time-dependent data in an efficient manner. [Goodchild M.F. 1988 p.318]

[b] Many minor nuances must be included when calculating the change between censuses:
Perhaps most fundamentally, the 1981 Census was taken on the night of 5 April (before Easter and out of term for higher education institutions) and the 1971 Census was taken on 25 April (after Easter and in term for higher educational institutions). In towns where the number of holiday-makers and students cause seasonal fluctuations in the size of the population, this three-week difference is likely to have some impact on the results obtained. [Norris P. & Mounsey H.M. 1983 p.276]

[c] Obtaining change information for high resolution mapping is difficult:
Finally, OPCS have not treated dealing with change at all seriously, at least compared to their treatment of the standard ‘snapshot’ census data. Additionally (and perhaps understandably in view of the problems of change statistics), little importance has been attached to local (less than district) level analyses. [McKee C.H. 1989 p.432]

71 [a] A policy of aggregating areas can prevent a proper study of geographical change:
The largest tract in the region, in terms of 1971 and 1981 EDs, occurred in the district of Bracknell in Berkshire, with 98 1981 EDs and 60 1971 EDs combining to form this comparable ‘small’ area (as defined by OPCS). This tract is therefore a good example of an area in which great change is taking place, but which — as a consequence — permits the least local study in the region of this change, due to the large size of the tract. [McKee C. 1989 p.4]

[b] Constituency boundaries are regularly redrawn:
The very poor correspondence between constituency and local government boundaries has arisen partly because the present constituency boundaries, which were first used in the February 1974 general election, were based on the ‘old’ set of local government areas. Normally constituency boundaries would coincide with counties and standard regions. However the current situation will continue until the next boundary revision is implemented. [Population Statistics Division, OPCS 1979 p.19]

[c] The approach adopted in this work could cope with the fine scale enumeration district study of census change over two decades:
Presumably, without comparable small areas and their related and comparable statistics, measuring local change between the 1981 and 1991 Censuses will necessitate the aggregation of EDs into comparable small areas between 1981 and 1991 by the user, where datasets permit. Comparison of change between the 1971, 1981 and 1991 Censuses to cover the 2 decades of change will, therefore, only be possible on a ‘supertract’ basis, by aggregating tracts comparable between 1971 and 1981 to areas comparable with the 1991 Census. This will reduce the amount of local detail considerably and, in areas of great change, such as the largest 1971/1981 tract in the region (made up of 98 1981 EDs) will certainly not facilitate a better understanding of change. [McKee C. 1989 p.19]

72 [a] The desire to aggregate as the time scale is extended must also be resisted:
Quite clearly the analysis of social change must incorporate a time dimension and quite often involves a complex interplay between several processes and between different levels of aggregation. Techniques to permit such complicated analysis are increasingly being developed and demographic research has become much richer through intelligent use of these approaches. [Hobcraft J. & Joshi H. 1989 p.11]

[b] Change can be very different at different scales:
It will be apparent ... that the relative significance of the different types of population change varies with the scale of analysis. Although decelerating increase is the most frequent type at regional, county and district level, accelerating decrease is just as common at county and district level but also occurs in three regions. In addition, the diversity of types increases with the reduction in scale of analysis, so that the pattern at district level is very much more complicated than at county and regional level; in particular the percentage frequency of types of population increase is much greater at district level than regional level. [Mounsey H. & Clarke J.I. 1981 p.6-7]

[c] Textual description of change can also be very elusive as a result of generalization:
The inadequately described have moved almost exclusively into the council sector. [Hamnett C. 1987 p.548]

73 [a] It should be possible to see the regional pattern through the local picture:
With the exception of the South East, in all regions containing a metropolitan county the balance of migration both in 1971 and in 1981 was outward; and in all the remaining regions it was inward. [Brant J. 1984 p.28]

[b] Again, contradictions are found in Inner London’s geography:
In central London in particular, however, things are not quite so clear-cut as groups of tracts with high percentage population increases lie side-by-side with groups of tracts which recorded high population decreases. [McKee C.H. 1989 p.198]

[c] Ward boundaries can be redrawn as often as every three years:
The most extreme examples are a new ward in the Isle of Dogs with a zero population in 1971 but 5,400 in 1981; and a ward in Bracknell district with a population of 3 in 1971 but 8,700 in 1981. [Craig J. 1988 p.9]

[d] Changing the scale changes the picture:
The City of London was the only London borough to increase in population during the 1970s yet it is precisely this district in which a number of tracts experienced some of the greatest decreases in population in the region during this period. [McKee C.H. 1989 p.201]

74 [a] The following is the first of three quotations which show how many times the same phenomenon can be independently recognised if it is clear enough:
The lowest mobility rate within a region was found in Wales (65); all other such rates were between 70 and 80. [Brant J. 1984 p.24]

[b] It is interesting to see how the numbers alter, but the picture remains the same:
The five districts with the lowest mobility rates per 1,000 total population are all in South Wales and include Afan (50) and Rhondda (56). Since many districts in this area also had particularly low rates in 1971 there seem to be local influences which reduce mobility as compared to other parts of the country. There may be a greater reluctance to move, associated with a strong community and family spirit, as well as problems in finding local employment and in obtaining accommodation. [Devis T. 1983 p.17-18]

[c] Slight spatial inconsistencies can still be found, however:
Rees (1978) has depicted the reluctance of population in one such community, the Upper Afan Valley in South Wales, to resort to labour migration. Interestingly, the valley’s neighbouring MELA, Rhondda, has the nation’s lowest per capita migration inflows and outflows. [Kennett S.R. 1983 p.223]

75 [a] The calculation of some census change variables is not totally reliable:
OPCS (1984) considered the change in the classifications of country of birth (factor H, appendix 3) to have “relatively little effect” on the measurement of change from 1971 to 1981. Even if this is so (and there are no means to prove it), a number of other factors — as appendix 4 indicates — affect the statistical comparability of these figures, one of which was a change in the editing procedures (factor I, appendix 3) from 1971 to 1981. OPCS (1984) considered that, in general, the effects of these would be lower than those caused by differences in the population base. [McKee C. 1989 p.11]

[b] Changes in migration can cause changes in the geography of voting:
For a long time this immigration was not regarded as being of any political significance, but as the 1950s passed the level of immigration and public unease increased. It was reflected at the national level by some members of the Conservative Party, and at the regional and local level by some Conservative Party associations and individual candidates in local elections. The most notable impact, however, was felt in the 1964 general election in Birmingham and the West Midlands. [Busteed M.A. 1975 p.49]

[c] While in the geography of employment:
The Black population of Britain is locked into an allocative system that seems bound to produce an increased polarization of native and immigrant populations. The forces that drew them into the economy are the same forces that are producing an increased isolation of the Black population. They came to fill gaps created by an upward mobility of the White population in the employment structure and they settled in gaps left in the urban structure by the outward geographical mobility of the White population. [Peach C. 1982 p.40]

76 [a] The simple view of a North/South divide is challenged by the spatial reality:
The use of unemployment as a convenient indicator of economic health is not new in this area of work. However, these studies have been given a new significance by two recent developments. The first arises from the growing realization that the simple division of the country into the “peripheral” slow-growing assisted areas contrasted with the more bouyant economics of the “core” areas of the South and the West Midlands is no longer tenable, and is unlikely to regain its former power in an era of rising national unemployment and slow or negative economic growth. [Frost M. & Spence N. 1981 p.7]

[b] Employment and unemployment levels are related, but not directly:
While employment trends in London have thus been dissimilar to those in other parts of the country, including the rest of southern England, changes in unemployment have tended to follow national trends rather closely, if with somewhat smaller fluctuations. The long-term decline in employment in London — which goes against the national trend — and the even sharper decline in inner areas have come about without a significant upward shift in London’s unemployment. [Buck N., Gordon I., Young K., Ermish J. & Mills L. 1986 p.180]

[c] The ward level shows us what is affecting individual people’s lives most precisely:
Analysis of trends over the 1980s points to a continuation of wide differences between the least and most privileged wards. Unemployment differentials have widened, even in the most recent period when the average level has fallen. [Congdon P. 1989 p.489-490]

[d] The doughnut which is clearly and repeatedly seen in the cartograms has been previously identified:
Particularly impressive is the “doughnut” of employment growth located at a radius of 50-140km from central London, which includes successful New and Expanded Towns such as Milton Keynes and Basingstoke, and Winchester, a county town which has gained from the growth of public services and from being at the centre of a region of rapid economic growth. [Champion A.G., Green A.E., Owen D.W., Ellin D.J. & Coombes M.G. 1987 p.63]

77 [a] Housing links the human geography of people to the physical geography of land:
Patterns of housing tenure in the conurbation are therefore a key element in the social geography of London and provide a social link between the built environment (the physical structure of London) and the social environment (the social structure of London in its spatial context). [Shepherd J., Westaway J. and Lee T. 1974 p.32]

[b] There is a debate as to how much polarization and segregation is due to the structure of housing provision:
By comparison of the standardised indices with the unstandardised indices ... it may be concluded that most of the segregation process operates through the housing market. The deviations from 1.00 (excepting the other group) do not exceed what one should expect from rounding errors. [Berge E. 1988 p.977]

[c] Some of the reasons for the collapse in the London housing market were foreseen:
Bramley and Paice (1987) have calculated that, even assuming that potential buyers can raise a 95 per cent mortgage on three times their income, one in three families living in the South East cannot afford to enter owner-occupation. [Hamnett C. 1989 p.111]

[d] Many people are forced to live where they do — while others can choose where to live:
A rational choice model is clearly oversimplified: the idea that couples have a free choice between sectors is — in both senses of the word — untenable: access to different types of housing is determined by “constraint” as well as “choice” (Rex and Moore, 1967). In Duesenberry’s words if “economics is all about how people make choices, sociology is all about why they don’t have any choices to make” (1960, p.233). For many, the idea of a free choice in housing is a sick joke, especially amongst the unemployed, those in insecure jobs, and for many in high house-price areas. [Murphy M. J. 1989 p.101]

[e] It is also interesting to see where most social housing is now required:
A much larger chunk of provision should be allocated to non-metropolitan areas in the south of England, and to housing associations and other agencies operating in these areas. Relatively little additional provision is warranted in the midlands and the north, areas which in the past have received substantial provision. [Bramley G. 1991 p.73]

78 [a] The first election shown in the series analysed here occurred in 1955:
It may be hard to believe nowadays, but during the 1955 general election campaign television news broadcasts made no references whatsoever to the election because the broadcasting authorities feared they would be in breach of laws regulating the conducts of elections. [Denver D. 1989 p.50]

[b] A great change can be seen in the image of February, 1974:
The February 1974 election was one of the most peculiar, and perhaps one of the most important, in British electoral history; it will be discussed more fully in the following section. It was called as a referendum on a specific policy issue for the first time since Stanley Baldwin did so — also unsuccessfully — over the tariff issue in 1923. As all know, the election not only stimulated the highest voter turnout since 1951, but also a mass exodus from both major parties — towards the Liberals in England and the Scottish Nationalists north of the Tweed. [Burnham W.D. 1978 p.280]

[c] The beginnings of transition were widely recognised at the time:
The electoral change in February 1974 was therefore quite exceptional, not simply in magnitude but also in direction: the British pendulum stopped swinging. [Crewe I., Särlvik B. & Alt J. 1977 p.132]

[d] The ratchet was also seen to slip in both 1959 and 1983:
The decline of loyalism within both major UK parties in the 1970s is well attested. Less obvious is the slip in support of 1959, linked to the resurgence of the Liberals in that year. [Budge I. and Farlie D.J. 1983 p.126]

[e] The implications of the changing geography of voting have been quickly realised:
A feature of voting behaviour in Britain in recent years has been the increasing volatility of the electorate, with a growing proportion prepared to shift allegiance between elections. [Johnston R. & Pattie C. 1989 p.241]

[f] Instability exists in a superficially increasingly stable electoral system:
If the socio-economic class cleavage basis of our present-day two-party system does in fact develop into a more even geographical distribution of partisan support, then in the last stage of the developmental model of bias the voter proportion distribution is symmetric and has a low variance. In terms of our beta-binomial model, intra-constituency variance gains in relative importance as the inter-constituency variance of the voter proportion distribution declines. The result will be large non-partisan ‘winner’s biases’ producing an increasingly less stable parliamentary system. [Gudgin G. & Taylor P.J. 1973 p.23]

79 [a] Only very simple electoral change can be shown on a single map:
Third, few maps have been constructed to show electoral trends on one map. This can only be done when constituency boundaries remain unchanged, but even when this condition is present too many results are simply presented as a series of maps for each election. For example, Adam failed to devise a map which showed electoral trends in France during the period 1947-62 even though the electoral boundaries remained unchanged. Trends in the pattern of election results can be shown by a bar-graph for each unit, where each bar represents one election and its length is proportionate to the percentage of votes obtained by the successful candidate. Electoral results in Belfast during the period 1920-1957 have been mapped this way. [Prescott J.R.V. 1969 p.382]

[b] The time series shown here is as long as that described below, the two-party decline does appear to have been of fundamental importance:
Until time-series data over a period similar to, and preferably longer than, that which forms the basis for this paper become available, the true impact of declining partisanship on the public legitimacy of the parliamentary system will remain unknown. [Crewe I., Särlvik B. & Alt J. 1977 p.188]

[c] Just as it is important to consider as many places as possible, times studied too should be numerous:
Fifth, it is important that conclusions should be based on the analysis of as many elections as possible. Reliance on a single election is dangerous because special issues or circumstances may produce an atypical voting pattern. [Prescott J.R.V. 1969 p.381]

[d] The tendency for contiguous constituencies to show similar behaviour is striking:
Spatial continuity has been a major characteristic of British electoral geography for more than fifty years (Johnston, 1983). It has continued since 1979, but with substantial and growing regional variations. The country was much more polarized in 1987 than in 1979: the London suburbs and exurbs, East Anglia, and the South West (though excluding Devon and Cornwall) have swung markedly to the Conservatives; the East Midlands and the nonconurbation parts of the West Midlands form an intermediate buffer zone of little change; and in relative terms the rest of Britain has moved to Labour. The Alliance, too, finds its greatest support largely, though not exclusively, in the areas of major Conservative strength. [Johnston R.J. & Pattie C.J. 1987 p.1012]

80 [a] Again, the North/South simplification should be avoided:
An interpretation of the May 1979 general election results as indicating the growth of a ‘two nations’ situation within England is an over-reaction to the simple cartographic pattern of changes in party allegiances. The analysis suggests that the inter-regional variation that occurred resulted from the consolidation of the strength of the Conservatives in regions where they did well in 1974 and relatively small shifts to them (relative, that is, to the national trend) in regions where Labour and/or Liberals performed well in 1974. [Johnston R. J. 1979 p.296-297]

[b] There has, however, been a regional polarization in some attitudes:
The Conservatives, in contrast, seem to be much more favourably placed: the pressure on the Conservative governments since 1979 to change their policies specifically to woo the industrial ‘north’ has been far less than that on the Labour party opposition to adjust its political manifesto in order to win support to the ‘south’. The Conservative electoral base in the ‘south and east’ and the Midlands — which together contain 60 per cent of British parliamentary constituencies — is large enough to assure the party political success. [Martin R. 1989 p.51]

[c] It is best to keep in mind the relative sizes of the respective populations:
The popular picture of urban and regional growth in Britain, in which the South is growing and the North declining, dates from the inter-war years but continues to exert a powerful influence. Today it is at best only a partial description of the truth, as Chapter 2 demonstrates. For example London, at the heart of the South, has lost well over half a million manufacturing jobs during the last twenty years. To put this into perspective, London has lost almost as many manufacturing jobs as Scotland ever had. Indeed, some of the fastest-growing areas are found outside the traditionally prosperous South East and Midlands. This picture is, in fact, one of considerable complexity. As a general rule the differences within regions are far greater than the contrasts between them. [Fothergill S. and Gudgin G. 1982 p.6]

[d] The growing divisions within London are stark:
Nevertheless, these trends suggest it is not impossible to envisage the development in the not too distant future of a socially polarized inner London, divided by tenure, with middle class owner-occupation juxtaposed with a residualized and predominantly working class public rented sector. Those groups excluded from this process may be displaced into outer London where many inner suburban areas may become transformed into lower value ownership mixed with much of what will be left of the private rented sector. The net result therefore will be a stabilising but polarized inner city and a declining suburban ring. In the process ‘inner city’ problems may become gradually displaced into the suburbs. [Hamnett C. & Randolph B. 1983 p.164]

[e] As the former capital of a long-gone empire London has suffered greatly:
In 1987 London’s economy is in deeper crisis than it has been in for a hundred years. In certain clear respects it is worse than the 1930s. The rigours of those depression years affected other regions more than London and did not bring quite the same extent of misery and insecurity to the capital. [Townsend P. with Corrigan P. & Kowarzik U. 1987 p.12]

[f] The second largest conurbation is often overshadowed by the first:
The West Midlands once again emerges as a black-spot where the cumulative human impact of recession was most severe, (particularly on this percentage measure: as a once-prosperous metropolitan region the scope for decline was greater here than in traditional problem areas such as Liverpool and Sunderland). [Green A.E. 1983 p.23]

[g] We must consider all these places and times together:
It would appear that the attempt to separate considerations of “regional structure” from those of “secular” and “historical” change can yield only partially valid results at best. [Duncan O.D., Cuzzort R.P., & Duncan B. 1961 p.174]