Segmentation and polarisation are not simple things to measure and there are no set ways of defining them. Take as an example the ten distributions just described in this chapter. In each case you need to go to the website of this book to get the underlying data which provides you with both the proportion of people allocated to each particular group in 1991 and how that proportion has changed in the years to 2001, allowing you to also calculate the number of people so allocated at the end of the period. The first website of the book, which used very differently presented maps, also has this data shown within each map, and it can be found at http://www.sasi.group.shef.ac.uk/hguk/chapter8.htm which is the web address for the old maps of chapter eight of this book.
If you sum the two percentages shown in each area on the old maps, the first versions of these maps to be drawn, then you will find the proportion working in that industry in 2001. All the figures in that on-line map, and in the nine which follow it, are proportions, expressed as percentages, of the entire population. They follow the same format to aid comparison. The versions shown in this book are simpler, just using four shades, but relying on exactly the same data. Whichever way you access it, you end up with ten sets of data, and each set containing two times 85 statistics. Although the numbers are only provided as whole percentages, without having this dataset on computer there are too many distributions to consider for any one person, therefore each select one of the ten distributions to study.
If there are 20 or more of you, you can each also select a year to study, either 1991 or 2001 (although the first and fourth methods defined below allow you to consider both years simultaneously). You need to determine how spatially polarised people were in the year you are looking at for the variable you are considering. There are many ways in which you can do this, some are listed below. Agree a method among yourselves, perhaps more than one, calculate the degree of polarisation which has occurred and then read on. Here are some methods you can use:
- Most simply, you can say a variable is polarising over time if the majority of areas fall into the shading categories: low and falling or high and rising. However, some variables are rising in every area or falling everywhere. In those cases you can subtract the national average (say median) change from the change measured first, and then redefine every place as either rising or falling in terms of national changes. Work out the median change simply by writing down the changes in order and selecting the middle, 43rd, one.
- You can measure the degree of segregation of a group at a given time. For instance, what proportion of people in the country would have to move between areas for that group to be evenly distributed across the United Kingdom? Or what proportion of the group would have to move, or what number of people would have to move, or what number or proportion would have to move if that group were to be distributed as everybody not in that group is! You have a lot of options as to how to calculate just these simple measures of segregation.
- You can measure the chances of someone chosen at random from the group you are studying meeting another person from the same group if that person were also chosen at random from within their area. This is called an isolation index and is easier to calculate than to describe (it is simply the sums of all the proportions weighted by those proportions, as you can assume each area has equal population). One problem for this index is that it tends to be higher the larger the proportion of a group is nationally. Can you correct for that?
- You can work out some average changes and draw a histogram of the results. For instance, in areas (grouped) which had a high proportion of people in a group in 1991, what has the average change in their numbers been? What about for average areas, below average areas and so on? For each type of area, as defined by proportion in 1991, there will have been an average change, which you can calculate. You can draw a histogram of those changes. If the tails of the histogram tend to rise and the centre falls, then polarisation has occurred. But what if the pattern is more complex than that?
- Think up your own way of measuring polarisation or segmentation and change in those measures. Ideally your measure should be simple to understand and preferably simple to calculate. It should measure something which is meaningful, its size should mean something and changes in its size should be readily interpretable. Can you think of a better way of describing whether the patterns shown in the maps above really do represent growing cleavages in the human geography of society or whether the changes are not as dramatic as that?
Having measured the levels of polarisation or segmentation in whichever way you have chosen, and having looked at the change over time in those measures, you next need to interpret the results. What has been going on? How would you explain your findings to an audience similar to yourselves? Can you appreciate why I have not included such measures here - or would their inclusion have altered the story being told in the chapter above? Should I have included such measures?
Having done all this work, what criticisms would you make of this chapter of the book? Can you tell simply from looking at the maps how they are changing? Is the United Kingdom becoming a more or less divided place, at least by what these measures show? And finally, why consider these measures? What really matters most about places in people's lives?
Related material for Chapter 8:
All available data and further material can be found in the Data section.