|
|
|
There is an alternative strategy possible to handle the problem of changing geographical boundaries for the 1961-91 panel file. In the cases where new districts have been created out of portions of two more old districts, a new hypothetical "super-district" can be created by aggregating together the old districts and the new district. For instance, in 1980, district Vizianagram (02 02) was created in Andhra Pradesh from parts of Srikakulam (02 01) and Vishakapatnam (02 03). In data from 1980 and after, the three districts can be combined into a single "district"; before 1980, Srikakulam and Vishakapatnam can be combined to produce the same geographic area. This is the method used in Bhalla and Tyagi, Patterns in Indian Agricultural Development.
This strategy has the advantage of producing exact matches over time without any estimation procedures dependent on assumptions of geographic homogeneity. It has the disadvantages of reducing the number of cases and creating extremely large "districts" that are far larger than the other cases in the sample. While our district files do not have these "super-districts", users can create them from our files by aggregating the appropriate districts (identified below). This is recommended as a check on the accuracy of the estimation procedures described above.
The following districts were created from portions of two or more previous districts. The percentages report the percentage of the 1961 population in the previous census that was incorporated in the new district; area percentages are reported in parentheses. More exact proportions are found in records 092 and 093. The changes are listed in approximate chronological order.
Recombining the following panel (version 5) districts will yield fewer districts but more exact over-time comparisons:
| Return to: Recreating 1991 districts before shifts in territory |
|
||
| Last updated October 1, 2000 |
comments to: Reeve Vanneman.
reeve@umd.edu
|