Geography Reference
In-Depth Information
Box 27.1 continued
Clearly, the Labour Party wished to get this changed
and presented an alternative scenario at the local inquiry,
whereas the Conservatives sought to convince the
Assistant Commissioner to recommend that the provisional
recommendations be retained as the final version. In
addition, two geographers presented independent evidence
promoting a set of constituencies that were more equal in
their electorates than that provisionally recommended, by
1980 electoral data if not those for 1976, which the
Commission was using. This last formed the basis for the
Assistant Commissioner's recommendation for a different
set of constituencies (Figure 27.6C), which was accepted
by the Commission (and in which Labour won five seats at
the next general election).
identified 13,317 (Johnston and Rossiter 1982), indicating
that the commission had a major task to identify that which
it felt best met the various criteria it was supposed to meet.
An important feature of the provisionally recommended
six constituencies is that, if how Sheffield's electors voted at
the most recent (1980) local government elections was a
reasonable guide to how they would vote in a general
election, then four of those six were likely to be won by Labour
and one by the Conservatives, compared with five for Labour
and only two for their opponents in the most recent (1979)
general election fought in the old constituencies. Statistically,
this was something of a surprise since, as the following table
shows, most of the possible 'solutions' would have retained
the
status quo
of a 5:1 Labour victory.
appointment of independent individuals or bodies
to produce more neutral configurations. In 1971,
geographer Dick Morrill was appointed to
produce a plan for Washington state without any
knowledge of what the two parties had proposed.
Table 27.1 shows the outcome for the 102-seat
lower house of the Washington State legislature.
In 1970, each party had won fifty-one seats, with
slightly more of the Republicans' than the
Democrats' being marginal. Each party's plans
Box 27.2
Redistricting Mississippi
The population of the State of Mississippi was around 40
per cent black in the 1960s, but none of the State's five
congressional districts defined in 1966 had a black
majority, the main area of black population (the Delta
region) being split between four of the five districts. This
'disenfranchisement' of the blacks was continued in the
State's districting plans after the 1970 and 1980
censuses, but the latter was challenged under the
Voting
Rights Act
in 1981 and a court created a black majority
district in the Delta region; this was largely retained after
the 1990 census, and the district has been won by a
black candidate since 1986.
What is the likelihood of a districting plan for
Mississippi including no black majority district, given the
size of the black population and its concentration in one
part of the State? This question was addressed using a
similar computer method to that employed by Johnston
and Rossiter (1982) in Sheffield. The 'building blocks'
used in American redistricting are much smaller than the
wards used in the UK, and although the equal size
criterion is much more rigorously applied, the number of
possible districting plans is very large—even if other
constraints, such as 'shapeliness' and not splitting
counties, are also applied. The procedure allowed only a
1 per cent maximum deviation around the average
population for the five districts, and generated 100,000
separate plans. In half of these, at least one district had
a black majority, and sixty-eight had two.
The implication is that a districting plan for Mississippi
without a black majority would be suspect, with the
boundaries having been drawn to prevent a black
representative being elected from the State (a classic
example of a negative gerrymander). But none of the
100,000 plans generated through the computer-intensive
method produced a district with as large a black
population as District 2 in Figure 27.7, suggesting that
blacks were 'packed in' to it; a larger black majority district
had been created by careful cartography than was likely
to occur by chance. Given that the adopted plan splits
more than three times as many counties as do plans
created by the 'County-conscious algorithms', the results
suggest that race was given undue precedence over
preservation of the integrity of political subdivisions.
In this case study, computer programs have been used,
not to generate a solution to a regionalisation problem,
therefore, but rather to provide a datum against which to
evaluate regionalisations. By creating a large frequency
distribution—a sample of all possible regionalisations in that
context—an 'objective' means of testing whether a
proposed regionalisation is 'unusual' has been provided.
