Normalized Differences:
Measuring normalized change per decade
This map shows the normalized differences for tracts over the course of a decade, from 1940 to 2000. Normalized differences is a good way to look at population change because it is based on a symmetrical distribution from -1 to 1. As Gregory and Ell stated the importance it in their work “Error-sensitive historical GIS,” normalized differences:
“[give] us a symmetrical measure from 2100.0%, complete depopulation of a previously populated area, to 100.0%, new population moving into a previously unpopulated area, with 0.0 indicating no change (Bracken and Martin 1995). Conventional rates, which just use ystart as the denominator, do not give symmetrical measures because while the maximum loss is 2100.0%, there is no theoretical maximum gain. Thus, using the normalized rates, a gain of 20% is the exact opposite of a loss of 20%, which is not the case with a conventional rate.”
This means that we can interpret deviations from the midpoint as being comparable to the same deviation in the opposite direction.
Overall, the 1940s and 1950s were periods of growth, whereas the 1970s and 1980s were periods of decline. The 1990s, however, show a much more healthy mix of growth and decline, and the locations are even quite surprising.
The 1940s are marked by stronger gains in the areas that were gaining population than there were losses where there was depopulation. The 1950s, however, showed losses in the inner city at a similar strength to the gains made, especially north of the city. The 1940s and 1950s stand out because of all of the rapid population growth in the ring of tracts that just escapes the reach of the “L.” This, as was mentioned before, is due in large part to the growing network of highways and the suburban boom. This same ring, however, starts taking a noticeable plunge in population starting in the 1970s and increasing in the 1980s.
The 1960s were driven by stronger losses in the city than gains in the suburbs. The 1970s and 1980s continued the same trend, with the only strong population change occurring in and around downtown where there were significant declines. For example, the three west bound “L” tracks pass through neighborhoods which experienced very eye grabbing decline from the 1960s through the 1980s. The people who were moving from here were likely contributing to the moderate growth in the suburbs.
The 1990s start showing population resurgence within the heart of the “L”s service area, in and around downtown Chicago. There is also some interesting growth around the Orange Line tracts down by Midway Airport. Large swaths of tracts around the northwest bound Blue Line to O’Hare have finally switched from perpetually losing population to finally gaining. Interestingly, directly to the south of the Blue Line is a collection of tracts which stand out as an area of current growth. This is also the first decade since the 1940s and 1950s that growth is more striking than loss across the region.
In an ideal world, the “L” would run through areas which have in recent years been growing. Even if an area is currently growing, however, it does not make up for the fact that the neighborhood might have been declining for many years. In what decades did the neighborhoods served by the “L” peak in population density? Is the “L” a system truly built for the neighborhoods of the past, or is it primed to step in to a role as the system of the future?
CLICK to see a map of the decades when each tract reached its peak density and see why this is problematic for the current state of the “L.”
Gregory, Ian N. and Paul S. Ell. “Error-sensitive historical GIS: Identifying areal interpolation errors in time-series data.” International Journal of Geographical Information Science, Vol. 20, No. 2, February 2006, 135-152. p.138