• 1852 Oversampling Strategy

"Donor" Sub-Districts

The 1852 Census of Upper and Lower Canada:
Proposed Oversampling Strategy, and Discussion

Selection of "Donor" Sub-Districts

Since characteristics varied from sub-district to sub-district, in part as a result of the timing of settlement, we would like to match each missing sub-district and division with a similar, geographically-proximate sub-district and division. To do so, we are entering into a spreadsheet information from the 1852 aggregate census statistics for districts which contain a missing sub-district. The information we are entering includes:

• Total population
• Birthplace (English, Scottish, Irish, French-Canada, U.S., other natives of Canada, other)
• Religion (Anglican, Presbyterian, Catholic, Methodist, Baptist, other)
• Sex ratio
• Number of families in the sub-district (to determine average family size per sub-district)
• Number of dwellings in the sub-district (to determine average dwelling size per sub-district)
• Male births in 1851
• Female births in 1851
• Males & females under 5 years
• Males & females 20-30 years by marital status
• Males & females 50-60 years by marital status
• Marital status by sex
• Total occupants of farm land
• Occupants of 50 to 100, 100-200 and 200+ acres of farmland
• Number of acres owned; under cultivation; sowed; in pasture; garden; woodland (to calculate % of land improved)
• Number of wheat mills; sawmills; carding mills; woollen mills; distilleries; tanneries; foundries; breweries; and the number of employees for each type (to roughly quantify industrial activity per sub-district)
• Number of dwellings classed as: stone; brick; wood; blocks; and shanty
• Number of shops; stores; inns; schools; public buildings
• (we cannot use occupations since they were only listed by district)

We plan to calculate percentage distributions of these socio-demographic characteristics by sub-district and division within districts, in the hopes of distinguishing suitable "donor" sub-districts. We would create a matching score and accept as a donor the sub-district with the highest matching score.

Resulting oversampled (or I should say extra-sampled) cases would be identified by a SAMPRULE variable, allowing us to determine the impact of including extra-sampled cases on the distribution of characteristics for each whole district and giving users the option of including or not including extra-sampled cases in their analyses.

We would like to be able to locate all sub-districts and divisions on a map; so far, we have found a map of 1852 census districts and sub-districts for the region of Montréal (20 districts) in Serge Courville, ed., Paroisses et Municipalités de la Région de Montréal au XIXe Siècle (1825-1861) Québec: Les Presses de l'Université Laval, 1988) and another map of Québec census districts in the Atlas Historique du Québec: Population et Territoire.

I welcome your feedback on this proposal

Lisa Dillon